最近更新

0.5 changes how you work in AudioSmith, not how it looks: colours, type and buttons are the same. You start from one box on Home; every tool starts its work at the top of its own page and lists what it made below; results hand off to the next tool; everything you make is in Assets and every long piece of work in Projects.

Where things are now

Before	Now
Studio	Text to speech
History	Assets
Voice gallery, My voices	Voices — Discover, Mine
Design a voice, Clone a voice	New voice — Design a voice, Quick clone
Library	Projects (search in it with the sidebar's search or ⌘K)
Models, Engine, Logs	Settings → Models & network, Engine, Logs
Integrations, Settings → Integrations, Local API	Developers — Overview, Local API (Settings → Integrations points there)
Tools	Developers → Tools
Activity cards at the sidebar's foot	the bell beside Settings
Top bar and breadcrumbs	each page's own header
Stories' and Audiobook's Project menu	the page's list, and the editor's ⋯

Old links (audiosmith://gallery, /history, /tools, /integrations…) open the new places; the Local API and MCP are unchanged. ⌘1…⌘8 are Home, Text to speech, Voices, Projects, Dub, Speech to text, Stories, Audiobook; ⌘F puts the keyboard in the sidebar's search.

Start from Home

• AudioSmith opens on Home: 「今天想做点什么？」 and a box. Type, paste a link or drop a file and press Enter (Shift-Enter for a new line): text opens in Text to speech as the script, a video or a link starts a dub, a recording is transcribed, a text file or an ebook becomes an audiobook. A row of tools and your recent work are under it, and five samples. With the writing help's model loaded, 「智能分派」 lets it pick between Text to speech, Stories and Audiobook.
• The sidebar lists places — Home, Voices, Projects, Assets — then the tools: Text to speech, New voice, Stories, Audiobook, Dub, Speech to text, Voice isolator, Dictation and ··· More (Workflows, Batch dubbing, Calls). The search is at its top (pages, actions, voices, projects, samples); Developers, Settings and the bell (tasks and notifications) are at its foot.
• Every page has its own header: its title, how many things it lists, and its own tools with at most one main button, all of one height. ⌘[ ⌘] and the mouse's back and forward buttons go through the pages you saw; an item's page shows its list's name before its title, and a click on it goes back to the list. In a narrow window the search becomes a magnifier and the less used tools go into ⋯.
• ⌘K searches everything. Empty, it offers quick actions and your latest work; typing groups pages, actions, voices (Enter uses one in Text to speech), projects, samples and takes by their script, and always offers to read what you typed in Text to speech.

Every tool starts on its page

• Text to speech: the script fills the left; chips under an empty script (Tell a story, Product narration, Podcast intro…) fill it with an example in the voice's language. Generate (⌘↩) is at its foot with the latest take. On the right, Settings (voice and model pickers, 「✦ Recommend」, speed, language, Advanced) and History (this page's takes, with Export, Reuse settings, Regenerate with another voice, Save as project, Delete).
• Dub: drop a video or a recording, or paste a link, pick the language and press 「配音」. It starts at once and shows on top of the list below with its progress; a click opens the dub's editor.
• Speech to text: drop, upload, link or record, choose the language, speakers and Fast or Accurate, and send. A transcript opens on its own page and hands off to 「配音」 and 「在文字转语音里朗读」.
• New: Voice isolator. Drop a song, a video or a noisy recording and get the voices alone, the music and effects, and the voices denoised, on this Mac (HTDemucs, DeepFilterNet); each track plays, saves and goes on to Speech to text or Dub.
• Stories and Audiobook open on a composer over your stories or books: describe a story or paste a script, or paste or import a manuscript and pick the narrator, then 「新建」. The project opens in the editor you know; the list's name in the header (or ⌘[) saves it and goes back to the list. Save and Save a copy moved into the editor's ⋯.
• New: 「帮我写」 on Stories. With the writing help's model on this Mac, turn it on and describe a story in a sentence: the writing help drafts a short scene for two or three voices in your description's language — written, without filler words — then the story is made, cast and opened. A progress line shows while it writes, with Cancel.

Voices

• 「发现 | 我的」 switch in the header.
• Discover: menus for language, gender, age and chips for use; 「Recommended」 cards of one width, then 「All」. ▶ plays in the row; a click opens the voice's details at the right: a large preview, 「Try a line」, and 「Use in Text to speech」, 「Use as a story character」, 「Narrate an audiobook」, 「Save to My voices」.
• Mine lists your voices with filters (All, Cloned, Designed, Saved); ⋯ renames, duplicates, exports or deletes (with Undo). A saved voice's details hold what the profile editor had.
• New voice offers Quick clone (a guided page: add a recording → check → name and save → done) and Design a voice (a description, preset chips, three candidates to pick from).

Projects, Assets and the samples

• Projects lists stories, audiobooks, dubs, transcripts and voice isolations, newest first, with type, date, length and state; search (titles, and the scripts of stories, books and transcripts), filter by type, table or cards; ⋯ opens, renames, duplicates or deletes. Isolations show their file's name, not its path.
• Assets gathers every output and every file you brought in — takes, renders, exports, isolated tracks, the recordings and videos you used — filtered by type and by tool. Each row plays inline, opens in its tool, and sends on to Speech to text, Dub, Voice isolator or (its text) Text to speech.
• Samples — 两人对话的故事, 一章有声书, 一段产品旁白, 一段配音短片, 一段会议转写 — on Home's 示例 tab, at the top of Projects and in ⌘K. Opening one copies it into your library as a normal project you can edit or delete, and opens it; opening it again asks whether to open your copy or make a new one. Each opens ready to play, and no feature introduction comes first: 「雨夜修表铺」, a scene for two characters and a narrator (the dialogue spoken by the multi-speaker model); 「北站的末班车」, a first chapter with two characters; thirty seconds of narration for a made-up lamp, as Text to speech's latest take; 「天空为什么是蓝色的」, a short English explainer of our own dubbed into Chinese, open in the dub editor; and a release meeting for two, transcribed with its speakers. All their voices are built-in ones.
• A dub copied or moved to another folder (a sample opened, the data folder moved in Settings → Storage) keeps its lines: editing one renders that line again, not every line. Dubs made before render each line once more the next time they run.

Help where you need it

• The first visit to Dub, Speech to text, Voice isolator, New voice, Stories and Audiobook shows a short introduction: three points and 「开始」. Each shows once; Settings → General → 「重新显示功能介绍」 shows them again.
• Empty lists show ghost rows that say what will be there.

Lists, panels and controls

• Every list — voices, projects, assets, dubs, transcripts, isolations, Text to speech's history, Home's recent work — has the same rows: no column titles and no lines between rows, rounded rows that tint under the pointer, while open and while playing, a click opens the item, ▶ and ⋯ at the row's end, and small icons before the length, date, type, language and use (their tooltips name them). Delete on a selected row asks to delete it.
• Details and tool panels float at the right of the page, under its header, sliding in and out: a voice's or a take's details no longer narrow the list; Text to speech's, a dub step's and a story's or book's panel keep their room beside the page while it is wide enough and float over it in a narrow window. ✕ or Escape closes them; a click on another row shows that one.
• Header filters, buttons and switches are as tall as the search field; Developers' and Text to speech's tabs are switches like Voices'.
• Play and pause are new icons (Hugeicons) everywhere audio plays; a preview shows pause, not stop, while it plays.
• Buttons and menu items no longer end in 「…」: Export, Import, Delete, Set voice, Settings and the others are just their names. Text that is still loading or waiting keeps its 「…」.
• Badges in a row — language, gender and age, use, a voice's kind, a project's type, an asset's source — are filters: a click shows only those, a second click clears it. ▶, 「Use」 and ⋯ in a row light up under the pointer, and every voice in Discover and Mine has a preview.
• Pages use a little more of the window: 16 points at each side instead of 24, the same for the header and the page. Home stays a centred column.
• 「··· More」 in the sidebar folds open in place (Workflows, Batch dubbing, Calls) and reads 「Less」; it opens by itself when one of those pages is opened from elsewhere.
• The menu bar icon shows a red dot while AudioSmith is generating (speech, dubs, renders, transcripts, dictation, the writing help) and a violet one while an agent is connected over MCP; downloads and waiting work don't count.

Dub editor

• The steps — transcribe, translate, voices, generate, export — are icons in the page's header: done ones are tinted, the current one stands out, running ones spin, failed ones are red; the tooltip names the step and its state.
• No footer: the step's main action sits in the transport row (or the header before there is a transcript), the state and progress ring in the header beside the panel button.
• The timeline: click the ruler to seek, drag the playhead, click a block to select its line (double-click plays from it); dragging empty space no longer scrubs. The lane is lower, and zoom, replay and ±5 s share the transport row. Space plays and pauses, Home goes to the start, J and L step back and forward.

Developers

• Developers opens on 「接入你的 Agent」: choose 编程代理, MCP or cURL and copy the snippet — a prompt for your coding agent, the MCP server's configuration, or a first request — then see 「等待第一次请求…」 turn into the time of the last request. Under it, links to the API reference, the skill and the Local API's settings, and every app and agent with its setup. The Local API and the tools (A/B voices, Convert, Rate fit, Direction, File probe, Watermark) are its other tabs.
• Agents can do everything the app does (except dictation, phone calls, the microphone, accepting licences and changing settings): the MCP server has 57 tools — new ones isolate voices, list assets and samples and open a sample, rename, duplicate, export and import voices, preview any voice, check a recording before cloning, give three design candidates, draft a story with the writing help, say a recording again in another voice, download a model, detect the watermark, probe files, suggest a delivery, check whether a line fits, and read the defaults. Dubs start from a link, transcripts take Fast or Accurate. The skill describes every tool with short recipes. Tested with ChatGPT (Codex) and WorkBuddy.
• Agents ask before they delete or download. A request that names the action is not taken as a yes: the agent says what goes or how big the download is and waits; clients that support it show AudioSmith's own confirmation. Tools say which ones only read, which delete and which go online. Agents are told to use the tools, not AudioSmith's window.

Fixes

• Stories: in a script without blank lines, a narration line after 「Name: …」 is narration again, not that character's line.
• A voice copy is named in the app's language.

Added

• The notch shelf, in place of the quick collect panel:
  • Rest the pointer on the MacBook's notch and a black shelf unfolds from it; it folds back as the pointer leaves. The menu bar icon and the quick collect shortcut open it with the keyboard (arrows, Return, Space, ⌘C, ⌘V, Esc).
  • The top row has the tools: capture an area, a window or the screen, record the screen, a voice memo, a link, paste, upload. Below are the items added last (scroll right for more): drag one out into any app, copy it, Quick Look it (Space), edit its name, tags, folders and note in place, or open it in Act.
  • Drag files, pictures, links or text onto the notch to collect them into the library open. What came in (or failed) is said under the notch, as the Dynamic Island does, instead of a notification.
  • Organize right after saving: once something is saved — on the notch, a screenshot, from the browser extension, a web address — the bar under the notch has small buttons to put it in a folder, tag it or add a note; each change is saved at once, and the bar folds away by itself.
  • Pin and tabs: 📌 keeps the shelf open, to drag one item after another into other apps. The + beside Recently Collected makes a tab of a folder, a tag or a smart folder, to drag its items straight out; each library has its own tabs, and the last one you looked at comes back.
  • Choose several, as in Finder (click, ⇧-click, ⌘-click, ⌘A), and drag, copy or Quick Look them together. Items in locked folders stay blurred and never leave.
  • Settings → Notch shelf, a page of its own with a live preview: opening on hover and its delay, dragging to the notch, which displays, a pill on displays without a notch (off as installed), showing it in screenshots, opacity, background blur, corner radius, and which tools the row shows, in what order.
• Watched folders ask on the notch: each library has its own watched folders (Settings → Library → Watched folders; Downloads is one click away). When a new download finishes, the notch unfolds with its thumbnail, name and size: Ignore, Save (copied into the library, the download left where it is) or Move (copied in, then the download goes to the Trash). Several downloads join into one prompt; after one, a new download within 2 minutes only shows a count beside the notch (the wait can be changed). Downloads still in progress, kinds Act does not take and files already in the library are not asked about. What is not answered waits, across restarts. Folders you watched before become the open library's and collect as before.
• AI Eraser (Tools): paint over clutter, a passer-by, text or a spot and let go — it is erased and what was behind it is made up, on this Mac (about 0.2 s an area). Hold ⇧ to paint several strokes first; [ and ] size the brush; Undo and Redo; hold "Original picture" (or ) to compare. Saved as a new version. The model (99 MB) downloads the first time.
• AI Upscale (Tools): make pictures 2× or 4× larger with detail a model draws in, on this Mac (a 1000 px photo at 4× in about 4 s). Photos, illustrations and anime, or fast; the original's format, PNG or JPEG; a new version by default, so the original stays. The preview compares a part of the original and the result with a line you drag, and warns when the picture has text, which AI may reshape. The model (about 37 MB) downloads from ActNow's own server the first time, fast from China too. Transparency is kept.
• Stacks: drag a card onto another and hold it there, or choose cards and press ⌘G, to pile them into one stack. A stack is one card in the grid, its cover on top and its count on it; choosing it chooses all its items (tags, folders, archive, delete, export, boards). Double-click to open it, reorder, set its cover or take items out; ⇧⌘G unstacks. Search still finds each item on its own, with Show its stack. Each item keeps its own tags, folders and rating.
• Archive: put items away without deleting them (the context menu, the bar of chosen items, ⌃⌘A). They leave All, folders, tags, smart folders, search and the counts, but keep their files, tags and folders, and are never cleared as the Trash is. Archive in the sidebar lists them, with Unarchive. Search finds them with "Include archived" or is:archived. Collecting something archived merges into it and offers Unarchive.
• Highlights and notes in texts: select words in a note, a Markdown or text file, a saved page's Reading view or a PDF's new Text view, and a small bar offers five colours, Note and Copy. The inspector lists an item's highlights (click one to go to it); their words and notes are searched, and has:highlight finds the items that have some. Highlights are found again after the text is edited.
• Change the date added and the source link in the inspector: put an old piece back where it belongs in time (one item or many; the original date can be restored), and add, change, copy or clear any item's source link. ⌘Z undoes either.
• Saving what you already have merges: saving a file or a page the library has makes nothing new and asks nothing — the item you have gets the new tags, folder and note, comes back from the Trash if it was there, and "Already in the library · tags merged" offers Open. Settings → Collect → When saving → "Ask each time" brings back the question. Results of a batch are said plainly: "Saved 3 items · 2 already in the library", with Details.
• Sites saved whole (with the browser extension): an X post (text, every picture at full size, its video, the quoted post, the author, the counts), a GitHub repository or issue / pull request, a YouTube or Bilibili video (title, channel, length, description, chapters), a 小红书 note, a Reddit or Hacker News post — each one item, shown in the new Card view, its text, author and chapters searchable. A product page's price is read and shown on its card, and can be changed.
• New in the browser extension (0.5.2; Safari's comes with Act — for Chrome, Edge and Arc export it again from Settings → Browser extension and click Reload in the browser's extensions page):
  • X: on a post's own page, Sync completes the item Act has — the whole text, every picture, the video, the quoted post — instead of saving a second one; your tags, folders, note and rating stay. The button shows ✓ for posts already saved.
  • GitHub: an "Act" button beside Star on a repository's page saves it as one card (description, homepage, language, license, stars, topics, README); once saved, Sync updates the numbers.
  • 小红书: a Save button under a note (beside "…") saves the whole note — title, text, topics, author, time and place, counts, every original picture or its video — as one item; the picture viewer has one too.
  • Pinterest: the Act button sits among Pinterest's own on pin cards and on a pin's page; video pins are downloaded as videos.
  • A saved link shows what is known at once: until its page is saved (or if it cannot be), a link shows a card with its title, site, author, length, description and a picture; bookmarks now send the page's details and a screenshot of the tab saved.
  • A clearer options page (what each site switch does, with pictures) and a tidier popup.
• Getting Started: a fresh install opens a short tour (six steps, about a minute): what Act is, your library and an Eagle library found to import, watching Downloads, the notch shelf played for real with sample cards (nothing written to your library), the browser extension, AI agents (MCP, Claude Code, Codex, Cursor) and the model pack. Help → Getting Started brings it back.
• Voice memos: record your voice in Act (⌘K, the File menu, the menu bar icon or the notch shelf); a small window starts, pauses and stops it, and the memo is saved as M4A and transcribed on this Mac, its words searchable. Up to 2 hours.
• Save Web Address (⌘K, the notch's Link tool) starts with the address of the page open in the front browser (Safari, Chrome, Arc, Edge, Brave, Vivaldi, Opera); macOS asks once to allow it.
• Open beside the items: ⌥↩, ⌥-double-click or Open Beside shows an item in a column right of the grid; the arrow keys change the item and the grid keeps its place. Settings → Browse can make every opening do so.
• Zoom the whole window: View → Make the interface larger / smaller / actual size (⌥⌘= ⌥⌘− ⌥⌘0), 80 % to 150 %.
• Put on a board from every item's page: videos, sounds, PDFs, links, posts, texts, documents and fonts as well as pictures, with recent boards and New board with this item. In a narrow window the page's buttons fold into ⋯ instead of pushing the name out.
• Free view while presenting a board (F): look about the whole board — drag, scroll, pinch to zoom, double-click an item to fly to it — without changing anything; Present goes back to the nearest page.
• Tools:
  • Installed tools that fit a picture or a video show in its page's header (the first five, the rest under ⋯).
  • The picture tools' dialogs show the original and the result side by side, with their pixels and sizes, as you change the form.
  • Compress: a custom quality (1–100). Convert: the sizes in the other formats. Resize: common preset sizes. Remove EXIF / location: says what it takes out.
  • Remove Background: Original, Crop to subject and Square in one click; the arrow keys or the wheel change the margin; it starts from your last settings.
  • Each tool keeps its newest 5 versions of an item, the older ones going to the Trash (Settings → Tools → Versions: 1, 3, 5, 10 or no limit); the original and the item's main file are never removed.
  • A tool that would leave less than 1 GB free on the disk says so and writes nothing.
• For AI agents: the act command line (search, read, add and organize items from a terminal, JSON out; Settings → Collect → MCP → Install the command line tool), and new MCP tools for the AI Eraser, AI Upscale, archive, highlights and the date added. What agents may do still follows MCP's switches.
• What's New: Help → What's New…, the command palette and Settings → About show each version's notes in Act's language; after an update, "New in this version" shows at the foot of the sidebar.
• Sidebar search: the magnifier becomes a clear button while there is text.

Changed

• Tags from what a page is about: saved videos, posts, repositories and articles get tags from their title, description and the site's keywords and topics, in the page's language, reusing your tags that mean the same — no longer just the site's name or the author. With the image understanding model, the local AI adds a few (at most 8 in all). The surest suggestions come first. The starter vocabulary has a new group of everyday topics (cycling, camping, hiking, cooking, programming…).
• Downloads and saves show with the other tasks: turning on collecting from links (downloading yt-dlp, about 55 MB), saving a post's video, a format extension's download and the like are tasks at the sidebar's foot, with progress and Cancel. Cards and the transcript panel say in one line what goes on, instead of a progress bar. "Download in the background" in the missing component dialog no longer starts the feature by itself minutes later: a notice offers Continue.
• Settings → Browser extension, its own page (⌘K: "extension", "Safari", "Chrome"): whether Safari has it on, the download page and guide for Chrome, Edge and Arc or export the copy Act holds, the connected browsers and what they may do, and its sound.
• The video tools' timeline (Cut a Clip, Extract Audio, Make a GIF, Compress Videos, Save a Frame) is tidier: frames and waveform in one rounded track, the selection a frame around it, a clearer playhead; the text under it is no longer covered.
• Lighter when idle and with large libraries: the notch shelf and watched folders wake Act far less while nothing happens; stacks, tags, similar items, highlights in long texts, long Markdown and boards with many lines are much faster.
• Smaller once installed: about 26 MB less disk (135 → 109 MB).
• Easier to read: text on the accent colour, on red buttons and on the notch shelf's blue is white and reads clearly in light and dark mode.
• The icons in the Tools menu are one colour that follows the menu, as other menus' icons.
• Settings → MCP says plainly that the agents' switches keep agents from making changes by mistake; they cannot stop other programs on this Mac.
• Libraries are upgraded the first time this version opens them (archive, stacks, tool versions): earlier versions of Act cannot open them afterwards and say "This library was opened by a newer version of Act. Update Act to open it." A copy from before each upgrade is kept in the library's backup folder. This version also adds settings an earlier one would drop: please do not go back to an earlier version.

Removed

• The media box. To keep a set of items at hand to drag into other apps, use the notch shelf's tabs. Its folder (~/Library/Application Support/Act/boxes/) is deleted a minute after Act starts.

Fixed

• Saved pages and links:
  • Behind a proxy app (Clash and the like), YouTube and other sites no longer fail as "addresses on this Mac or the local network are not fetched".
  • A saved Bilibili page no longer stays black on its item page; a live page that needs scripts says so, with Run scripts.
  • Why a page or a post's video was not saved is said in plain words, never an internal code, with Try again. When collecting from links is off, a post's card offers Turn on and save the video, and Settings → Collect can save all such videos at once.
  • An X post's video and avatar are no longer lost; posts saved before can be completed with Sync.
  • Links analysed before their page came in are analysed again once it is there (also for links saved before, once, in the background).
  • Suggested tags no longer say what a page is in rather than what it is about ("中文", "字幕", "录屏", YouTube's "免费", "上传"…), unless you use one on 3 items or more.
  • Bookmarks can be found by meaning and get folder suggestions.
• The item's page:
  • Turning to a post, a text, a table or a font shows it as it will look, without a jump or a blank; neighbouring pages are read ahead.
  • On saved pages and PDFs, pinching in closes the page as over a picture; PDFs and the live page zoom when pinched; scrolling to the end and swiping again turns to the next item, and no longer too early.
  • No black line under web views and no black past a page's edges.
  • The first two-finger swipe after a pause turns the page, and the first scrub of a video goes on.
  • A star rating (or any change) shows at once, not after the second click.
• The notch shelf: nothing shows outside it or a light fringe around it as it opens and closes; the download count beside the notch opens it when hovered; the count is not shown twice; cards' rings no longer shrink their thumbnails; no empty band after a paste or a drop; long pictures are sharp.
• Choosing and undo: ⇧-click keeps what ⌘-click chose, as in Finder, in the grid and on the notch shelf; ⇧-click and the keyboard stay on their items when new ones come in at the top; ⌘Z in the grid undoes the latest change.
• Large and odd files: deeply nested JSON, very wide Markdown tables and malformed packs no longer use up memory; the AI tools and collages check a picture's size before reading it, use far less memory on large pictures, take turns instead of running out of memory, and say plainly when a picture is too large.
• Disk and storage:
  • Settings → Storage → Clear clears the caches Act really keeps and says what it freed (it freed 0 bytes).
  • Leftovers of a crash or a forced quit are cleaned up.
  • The daily backup of a library's index is instant on APFS and takes no extra space until it changes.
  • Screen recording does not start with less than 2 GB free, stops and saves before the disk fills, and keeps what was recorded before a crash.
  • A full disk is said as "The disk is full. Free some space and try again."
• Locked folders keep more shut: third-party tools, agents, the local API, exported packs and the menu bar's Recently Collected no longer reveal or change items in locked folders.
• Watched folders take only the file itself (not a link put in its place), and Move sends a download to the Trash only when the library has its own unchanged copy.
• Downloads and watched folders go into the library they were started for, not the window in front.
• A library from someone else that you have not trusted no longer links or reads files outside it.
• Saving a linked text writes over your own file, keeping its Finder tags, comments and aliases.
• Model downloads try again when a proxy drops the connection, and say what to do when they fail.
• Act's window shows at once on start (it waited about 1.7 s).
• Boards: an agent's change and yours made at the same time no longer undo each other.

Security

• Pages and links never reach your Mac or local network: addresses are checked where Act connects (DNS rebinding no longer gets through), with more local address ranges; saving a page renders public pages only. Video downloads check their address too, refuse live streams and stop past 4 GB or 6 hours of media. A download that sends nothing for 60 s ends.
• The clipboard card no longer opens copied links (a password reset or a signed download link): it shows only the host and the site's icon. Insert takes only what the card offered.
• The browser extension's pairing token cannot add files of your Mac, nor tell whether one exists.
• Model downloads: every file of every model is checked against the checksum Act ships.
• Updates are signed for their version and channel (stable apps take no beta) and come over HTTPS only.
• Previews of saved pages and documents are sandboxed: a page cannot take the preview elsewhere, links open in your browser, nothing is downloaded unasked.
• Dropped pictures: a link, a page or a program dropped as a "picture" is not taken.

Added

• Reply tracking on a mail you sent: a button right of Forward (in Sent, or any conversation you wrote in) sets "Remind me if no reply" for it after the fact, with the same presets; it shows the setting in force and can change or stop it. Replies that came before it was set are not waited for again.

• Snoozed conversations can be ignored (I, the context menu, More's "Ignore all", the bar): the snooze is cleared, so they leave Snoozed and do not come back at the snooze time; ⌘Z puts the snooze back.

• Images in mail: insert a picture into the body at the cursor (also from the right-click menu), resize it by its corners with the aspect ratio kept, and it goes out as an inline picture.

• Attachment thumbnails: pictures, videos, PDFs and plain text show a small preview in the message; click for a larger view. Archives and other files keep the plain card.

• AI subject: a ✨ button in front of the subject appears once the mail has text; it writes a subject from the body (undo with ⌘Z; a reply keeps "Re:").

• Right-click in the body: cut, copy, paste, paste as plain text, select all, undo, redo, and the AI actions (draft, shorter, friendlier, more formal, fix spelling, translate).

• Compose type: a larger body (15 pt, 1.6 line height) and pasted or typed web addresses become links.

Fixed

• Translation in the reader now covers every message in a conversation, not only the newest one: collapsed messages are translated too, and their summary lines follow the translation. A translation that was interrupted (switching back to the original or to another language) is completed the next time.
• Time and star on each message in the reader line up on one centre line.
• Home: a card dropped after pausing over another card stays where you put it.
• Home: the default layout is the nine cards of the first release: today's focus, brief, to reply, deadlines and meetings, overview, incoming trend, subscription cleanup, privacy and security, and accounts.
• Drafts open at once: a draft's body comes from this Mac when it is here, instead of being downloaded every time.
• Background work is quiet: indexing, prefetch and attachment reading no longer pop up cards at the bottom left.
• Sent copies: a copy that cannot be saved to Sent is retried, and each result is logged.
• Translation shows paragraph by paragraph, with a count while it runs, and no longer flickers; a partly translated mail is reported as not finished.

Dub timeline: zoom and pan

• Zoom the waveform timeline with a trackpad pinch, ⌘ or ⌥ with the scroll wheel, or ⌘= and ⌘−, around the pointer; ⌘0 or a double click on the zoom readout shows it all. It zooms from the whole source to 100 pixels a second.
• Pan with a sideways two-finger swipe (with the system's inertia), ⇧ with the wheel, a middle-button drag, or the new overview strip under the zoom controls, which shows the part in view over the whole source; drag it, or click beside it to jump there.
• While playing zoomed in, the view turns the page as the playhead reaches its edge, unless you panned in the last few seconds.
• Dense lines no longer crowd the timeline with numbers: a number shows only where it fits and leaves room after the last one.

Dub: see the video before it downloads

• Paste a link and the Dub page shows what it is within seconds, before anything downloads: the cover, the site (YouTube, Bilibili, Vimeo, TikTok and the others yt-dlp knows), the title, channel, length, picture size, estimated download size, date, views, the description (Expand shows all of it) and the caption languages. It is kept in the dub, so it shows again without the network.
• While the video downloads, a thin bar under the cover gives the size and time left; once it is downloaded, the centre shows the video itself, with play, while transcribing and the steps after it report in the footer.
• A link that cannot be read (a private or removed video, one that needs a sign-in, a site that is not supported) says so before any download, with the site's reason, and Try again. Cancel stops the reading as it stops a download.
• A dropped video or audio file shows its poster, length, picture and size the same way.
• The cover is fetched by the app itself, through the same proxy as the download, from public addresses only, at most 5 MB, and kept only if it is a JPEG, PNG or WebP picture.

Faster with long projects

• Dub: the editor builds only the lines in view; a 2-hour video's 2,000 lines no longer slow every frame (190 ms → under 1 ms a frame).
• Audiobook and Stories: a long manuscript (a novel) is checked and its draft saved once you pause typing, away from the window, not at every key; typing in 500,000 characters is as quick as in a short text.
• Studio and Audiobook editors: wrapping long text remembers what it measured, so opening, resizing and typing in a long manuscript no longer stall (rows still never pass the editor's edge).
• The invisible watermark is added while the audio streams through, so watermarking an hours-long dub no longer holds the whole recording in memory several times.
• The engine keeps up to 6 models loaded (small ones don't count) and lets the memory budget decide, so a story render no longer reloads a model at every line.
• Finished tasks older than 30 days are removed at startup; voice previews no longer copy their audio every frame; the render cache is tidied at most once a minute.

Security

• Voice names are cleaned wherever they come in (imported .asvoice and .ovsvoice files, the Local API, MCP, the app, Discover voices): line breaks and control characters become spaces, and a name keeps at most 80 characters. Names saved earlier are cleaned the next time they are saved.
• The agents' skill and setup prompt list voice names as data (quoted, cut short), not as text an agent could take for instructions.
• MCP voice cloning and design work only from this Mac (or with the API key), like the Local API's; over the LAN, a voice's audio is its sample, never the reference recording.
• Built with Go 1.27.3 (security fixes in Go's standard library).

Logs

• An operations log in the app's log (source “ops”, one key=value line each): creating, renaming, deleting, importing and exporting voices, projects, dubs, transcripts, workflows and takes, with who asked (the app, the Local API, LAN, MCP); settings changes by name; each job's start, end, failure (with its error) and cancel (with who cancelled and how long it ran); downloads; Local API and MCP calls that change something, with their status; errors shown as toasts; the app's start and each step of its quit. Never script text, transcripts, prompts or secrets.
• Logs and diagnostic bundles mask Local API keys, Twilio tokens, keys read from the Keychain and bearer tokens, and show the home folder as “~”; the engines' own log files are masked too.
• Errors that were dropped silently in background work (cancelling jobs, export records, notifications, files the app writes or removes…) are logged.

Fixes

• Studio and Audiobook script editors: the last character of a line was cut in half at the right edge on the Mac (a Chinese comma or full stop measured half as wide on its own as within its line). Lines now wrap where they are drawn.
• Importing a video from a link shows its real progress (percent, size, time left) on the page, in the footer and on the sidebar card, with the video's title; Stop on the card and Cancel in the footer really stop it (yt-dlp's helper process kept downloading before), and activity cards no longer show an empty bar.
• A copy built on the Mac says, in Settings → About → Software updates, why it cannot update itself, with a link to the download page.
• My voices: Delete… (and the other items) in a voice card's ⋯ menu did nothing — no dialog, no log line — unless the voice's editor was open: the menu went away with the pointer before the item chosen reached it.
• Deleting a voice says where it is used (stories, books, dubs, workflows, the Studio, batch dubbing, calls, agents) and can be undone from its toast for about 10 seconds; its files wait in the library's trash until then. Places that named it show “Voice deleted” to choose another one, a story or book that uses it stops with a clear message, and the Studio says its voice went — never another voice in silence.
• Quitting never hangs: edits not saved yet (drafts, the dub editor's changes) are written first, and jobs that do not stop when cancelled — such as a stuck download — are left after a few seconds (they continue or wait for Retry at the next start). An engine that stops reading its requests is restarted instead of blocking.
• A job after an interrupted one (the engine stopped twice in a stage) no longer waits forever: it fails, saying why, the dub or book is no longer shown as busy, and Resume continues it. Chains stuck this way in an earlier version recover at the start.
• The language model's turn on the GPU lasts until its request has really stopped, so the next job is not refused as busy.
• Models moved by a move that did not finish (cancelled, a model in use by another app, the app quit) no longer show as not downloaded; the move continues after a restart.
• Deleting a dub waits for its running stage to stop; two runs of the same dub started at once make one.
• Stories and Audiobook: a rename on the Projects page no longer marks unsaved changes as saved; saving over a project changed elsewhere (by an agent through the Local API) asks whether to save a copy or replace it, and the Local API refuses an edit made on an outdated read.
• Moving the data folder measures it and cleans up a failed copy in the background, with the dialog saying so, instead of freezing the window.
• Smaller fixes: a crash in background work is logged instead of closing the app; a draft that could not be written is tried again; unfinished temporary files are cleaned in every folder; downloads left running by a crash are stopped at the next start.

Inboxly's first beta: every mail account in one inbox on your Mac — sorted for you, searched by meaning, summarised by models that run on the Mac. For macOS 15 or later on Apple silicon.

Added

• Every account in one inbox: Gmail, iCloud, Yahoo, AOL, Fastmail, Zoho, GMX, Proton Mail (Bridge), QQ Mail, NetEase 163/126, Aliyun Mail, Tencent Exmail or any IMAP server, added with an app password or 授权码 and each provider's steps. All, Primary, Updates and Promotions, To reply, Snoozed, Starred, Drafts, Sent, Archive and your folders. New mail arrives as it comes; when the network, the Wi-Fi, a VPN or the proxy changes, Inboxly tries again at once, and when it can't connect it says why — a proxy that blocks the mail ports included, with what to change.
• Fast with the keyboard: single keys (E archive, S snooze, # delete, J/K…) or Gmail's, ⌘K for every command, page and search, and Undo (⌘Z) for what you do to your mail.
• Safe reading: HTML mail without scripts or trackers; pictures load by themselves only from trusted senders; phishing warnings with their reasons, the sender's SPF, DKIM and DMARC, and risky attachments blocked or asked about.
• Sorted for you: Primary, Updates and Promotions; To reply for mail that waits for your answer; tags for deadlines, meetings, sign-in codes and cold outreach; your own categories and rules; Subscriptions with one-click Unsubscribe; blocked senders.
• Search by meaning: keywords at once, mail close in meaning a moment later, each result saying why it matched; from:, has:attachment and after:; search on the server; similar mail.
• AI on your Mac: a summary of each conversation with what's asked of you, suggested replies, Draft with AI, Rewrite, translation, and Ask your mailbox, answered with links to the mail — with the ActNow family's models, which a Mac with Act has already. Nothing leaves the Mac.
• Writing: rich text or Markdown, recipients completed as you type (pinyin too), signatures, templates, attachments checked against the server's limit; undo send, Send later, the outbox and drafts kept on the server; the mail looks the same in Gmail, Outlook, Apple Mail and QQ Mail.
• Later: snooze, Remind me if no reply, and a helper that sends scheduled mail and brings back snoozed mail while Inboxly is closed (Settings › Composing).
• On the Mac: notifications for people's mail, with Reply, Archive and Snooze; unread mail in the menu bar and on the Dock; quick compose from any app (⌃⌥⌘N); Share to Inboxly; desktop widgets; Spotlight, Calendar, Reminders and Contacts once you allow them; Touch ID lock; the default mail app if you like.
• Home: today's focus, today's brief, deadlines and meetings, subscription cleanup.
• Activation and updates: activate Inboxly with a licence code from actnow.dev; Settings › Licence shows the code, its end date and the Macs in use, and Inboxly works offline for up to 30 days between checks. A new version shows at the foot of the sidebar with its notes in your language and installs when you say so (Settings › About › Updates).
• Getting started and help: a welcome with sample mail to try before adding an account; Help › Send Diagnostics with addresses masked.

Act's first version: a home on your Mac for the images, videos, audio, documents and web pages you collect — kept in libraries you own, found by what they show and say, understood by models that run on the Mac.

Added

• Libraries: .actlib folders you own, with every item's data beside its file (Repair rebuilds the index from them), daily backups, a 30-day trash; several libraries, switched from the sidebar. Files are copied in (cloned on the same disk) or linked where they are: linked files are watched, found again when renamed or moved, marked missing when gone (Locate…), and read again when edited.
• Collecting: drag and drop, paste, Import files…, links (pages, images, videos through yt-dlp), screenshots, a Quick Collect panel and a menu bar icon with global shortcuts (Quick Collect on ⌃⌥⌘S, off until you turn it on), watched folders; duplicates are noticed before they are added.
• Browsing: a grid, justified rows, a masonry and a list (⌥1–⌥4), fast with 10 000 items; quick look (Space), an item's page with its strip, ⌘O to open an item in a window of its own (⇧T keeps it on top); previews of pictures (zoom, 1:1 of very large ones by region), GIF/APNG, SVG, videos and sounds (converted on the fly when WebKit cannot play them), PDF with a page strip, text and Markdown, tables (xlsx, csv) and documents (docx, rtf); the inspector with tags, folders, rating, note, source, file details and versions.
• Organizing: folders with icons and colours, smart folders with a rule editor, the tag manager (groups, starred tags, merging), Quick access, batch rename (templates, find and replace with regular expressions, undo), duplicates and look-alikes with a merge that keeps tags and folders, annotations on a part of a picture or a moment of a video.
• Search: the search field finds by names, tags, notes, text in pictures (OCR), documents' contents, what is said in videos, annotations and meaning (EmbeddingGemma 2 on the Mac), with filters (type: tag: ext: rating: added: …) and colours; saved filters; ⌘J to jump to a folder or tag.
• AI on the Mac: a pipeline that reads pictures (Vision), transcribes speech, embeds everything for search, suggests tags, folders, titles and file names; To review collects the suggestions; it pauses on battery, heat or memory pressure, retries what failed later, and needs no account or cloud.
• Import from Eagle: libraries with their folders, smart folders, tags, ratings, notes, links and comments, copied or linked, synced again incrementally; Eagle's .eaglepack packs too.
• Packs: export items or a folder as an .actpack (with their metadata, annotations, versions and attachments) and import one into any library.
• Web pages: saving a link keeps its page — a snapshot, its first screen, its readable text — and a browser extension (Chrome, Edge, Safari) saves images, pages and links straight into Act.
• Saved links' covers and previews: a link's cover shows as soon as its page has loaded (its first screen; the page's own picture when the first screen is blank, a video's own picture), before the rest is saved; one that cannot be fetched says so on its card and is tried again later, less often each time. Links from Eagle (with Eagle's screenshot as their cover until then) or saved before get their cover in the background, one at a time (Settings → Collect). The inspector switches the cover between the first screen and the page's picture, and an item's page makes any part of the whole page the cover. Cards show the site's icon and domain. Space opens on the cover at once, then the live page — without its scripts unless you run them — or the article; links clicked in a preview open in the browser.
• For other apps and agents: a local API with a token, an MCP server (Claude Code, Codex, Cursor…) with a Claude Code skill, and act:// links.
• Third-party plugins (developer preview): inspector panels, windows, actions and formats others write, each running with only the permissions you granted (actions in a sandbox); off until you allow them in Settings → Tools.
• Several libraries at once: each library in a window of its own (⌘N, Open in a new window); the windows open again where they were at the next start, and what you collect from outside goes to the front window's library. Copy or move items between libraries — drag them to the other window, or Copy to library › / Move to library › — with their tags, folders, rating, note, source, annotations, versions and attachments; content a library has already is not copied twice.
• Sharing: Share › in the items' menu and the Share button hand their files (a linked item's where it is) to AirDrop, Mail, Messages and the Mac's other services.
• Fonts: TTF, OTF, TTC, WOFF and WOFF2 show an "Aa" thumbnail and a sample page — your own sample text at any size, the sizes under it, every character the font has — and their name, family, weight and glyphs in the inspector.
• Actions: save a run of steps — add or remove tags, add to or move to a folder, set a rating, run a tool with its settings, export to a folder — in a library, and run it on the items chosen from their menu, ⌘K or ⌃⌥1–⌃⌥9; ⌃⌥Z undoes a run (not the versions tools made, nor the files exported).
• Updates inside Act: signed updates on the beta channel, as a small patch when one applies, the full package otherwise. A daily check shows a new version at the foot of the sidebar; its panel has the notes in your language, downloads with progress and restarts to finish when you say so. Settings → About lists every version's notes and has "check automatically". Early-access users are told before an update that needs a paid licence, and can keep the version they have.
• The family's act:// links: other ActNow apps hand files to Act with their tags, note and source (act://import); paths that are hidden, inside apps or libraries, or not on the Mac are refused, and Act lists what a link hands over and asks before collecting it. Act tells the family which models it has (search, speech, image understanding); models are kept in ~/Library/Application Support/ActNow/models, those already in the Hugging Face cache are brought over without downloading them again, and Time Machine leaves models out unless you turn backups on in Settings.
• Transcripts and chapters: a video's or a sound's page shows its transcript beside the player — the line playing follows along, a click goes there, a line you correct in place is searched at once — and exports it as SRT, VTT, text or Markdown, or saves the subtitles as an item; chapters are cut by the pauses and titled by the local model, marked on the progress bar; word timing on request.
• Ask AI: ask the local model about the items chosen (four questions ready, or your own); the answer streams in, and can be kept as a note or a Markdown item.
• Search by picture: drop, paste or choose a picture (a file, an image from the browser, a link) in the search field to find the items that look like it and those about the same thing.
• Document summaries: PDFs, Office files, texts and saved pages get a suggested summary (it becomes the description) and keywords as tag suggestions, in To review; file details (camera, lens, the site a picture came from) become tag suggestions too.
• Manual order in folders: drag items (several at once) where you want them; the order is kept with the library and comes from Eagle.
• Small things that help: sounds play while the pointer rests on them (Settings → Browse); a ✦ on items with suggestions to review; a folder's subfolders as cards above its items, with covers you choose; ⌘F finds words in a PDF and goes to their page; pictures zoom with the trackpad as in Preview (pinch, smart zoom, pan); letters jump to items by name; R turns a picture shown.
• Keyboard: change the window's shortcuts (Settings → Shortcuts: record, conflicts, restore), the menu bar lists its commands with their keys, the ? sheet lists every key and searches them; Esc, the arrows, Space and Return work alike in the grid, the quick look and an item's page.
• Repair and restore: repair a library with progress (backed up first, cancellable) and restore it from one of its backups, from Settings → Library or the library menu.
• Safari: the browser extension comes inside Act for Safari (turn it on from Settings → Collect).
• AI throughput: pictures are embedded before they are read, so search by meaning covers a new import sooner (text in pictures comes a little later); ACT_AI_TIMING=1 logs the time each stage of the AI pipeline takes.
• Right-click menus: grouped by what you do most — open, copy, organize, find, share and export, AI — with icons and the keys you set; what is seldom used under More ›; the view's empty space has a menu of its own (import, paste, sort, layout, thumbnail background, slideshow, export as CSV).
• From Eagle's menus: pin items to the top of every view; duplicate (⌘D); a new folder with the items chosen (⌥⌘N); replace an item's file (the old one stays as a version); your own thumbnail (a picture, the clipboard's, or a video's frame); a full-screen slideshow (⌥⌘Y); export as CSV; combine pictures (the Collage tool).
• The window: no status bar — what is going on (AI analysis, imports, downloads, a stopped engine) shows as cards above the sidebar's icons, with pause and cancel; Settings, Tools, the notifications and the logs are icons at the sidebar's foot; the models are in Settings → Models and network; Settings → About has the version, updates and licences.
• Duplicates: Check existing items finds the same files and the pictures that look alike across the whole library in the background (100 000 items in about a minute), without the AI models.
• Viewing pictures: zoom as in Preview — pinch around the fingers, momentum, a double click or a two-finger double tap for 100 %, ⌘ + scroll — with the zoom shown as it changes; go through the items with ← →, a three-finger swipe or a two-finger horizontal scroll; Find similar (⇧⌘S) from an item's page, the quick look or the inspector.
• Colours: the palette keeps small vivid accents (a red logo on white) and takes its colours from the picture's own pixels; a video's from its shots.
• More fonts: collections and dfont files show every face, variable fonts their styles and axes; install a font from its page.
• Big libraries: scrolling keeps in memory about what the grid shows, and importing writes the items' data faster.
• Part of the actnow family: shared settings, model folder and memory budget with the other apps; light and dark appearance, English and Chinese.
• Icon: Act's own icon — a masonry wall of white cards whose fourth cell is a sky-blue dot, on deep oxblood — and the same mark, one weight lighter, for the menu bar (the actnow family's icon style). The cards have rounder corners (Dock, Finder and the menu bar). The menu bar icon's dot turns blue when something wants a look — a notification, new suggestions to review, work finished in the background — until Act is opened (Settings → General can turn it off).
• Sounds (Settings → General → Sounds, off by default): a sound when a task finishes (a model download) and another when one fails, each chosen from the ActNow family's sound library (Bloom and Shrug by default), macOS's system sounds or none, with a compact picker that searches them (names and tags), filters by type and plays each with ▶ (again to stop); a volume in steps.
• Activation: Act asks for the licence code from your email or your ActNow account the first time it opens (early access: request it on actnow.dev). The licence works offline and checks in with actnow.dev once a week; Settings → Licence shows its edition, when it ends and the Macs it is on, opens actnow.dev to manage them, and changes the code. Until Act is activated, the menu bar icon, the global shortcuts, watched folders, capture, the local API and act:// links wait; when a licence ends, your libraries and settings stay as they are.
• Nothing missing stops a click: a tool not installed or turned off, a model not downloaded, ffmpeg not set up — clicking what needs it shows what it is (its size, where it comes from, what it may do) with "Install and continue", "Turn on and continue" or "Download and continue", and what you clicked goes on with the same items once it is there (a long download can go on in the background); Cancel does nothing. Error messages say what happened in your language (the details go to the log). Agents get a code such as tool_not_installed with the tool's id.
• Save X posts: every post on X (the timeline, a post's page, replies) has an Act button left of its bookmark button, the same size and colour as X's own; it saves the whole post — a text post as a text item, each picture at its original size, a GIF's video, a video's post for Act to download — with the text as the description and the post's address, author, their page and time as the source (the inspector shows them and opens the author's page). A post saved before shows ✓ and is not saved twice. A Pinterest pin's page shows one "Save to Act" over its picture again (two showed, one above the other), clear of the page's back button.
• X link cards: a post with a link card adds the card's title and address (the real one behind t.co when it can be told) to the end of its description, or of its text for a text post.

Fixed

• Saved links' covers and titles: when a page's picture (its og:image) or icon turns out not to be a picture — an address that redirects to the site's home page, or a page a server says is an image — it is no longer kept and can no longer become the link's cover; real pictures a server mislabels (WebP, HEIC, AVIF, SVG…) are still kept. A page whose title is only the site's name (Figma's files are all titled "Figma") takes the more specific title it gives elsewhere (twitter:title or <title>).

Security

• Libraries from someone else: a library's linked items can name any file of your Mac. Opening a library Act has not seen that names files outside it now asks first — "This library refers to N files outside it", with where they are (folders of keys, passwords or other apps' data marked) — and until you trust it Act does not read those files: they are not previewed, analysed, packed, nor given to the local API, MCP or the AI. Your own libraries (made on this Mac, or opened before this version) are trusted without asking; Settings → Library shows the answer and changes it.
• Packs leave linked files out: exporting a pack with linked items asks first; their files go into the pack only when you check "Include the linked files", which lists where they are. The items left out keep their names, tags and notes in the pack, and importing it gives them to an item with the same content.

The window in the ActNow family's layout

• The sidebar starts with a search field (⌘F): it searches projects, voices, takes, transcripts and exports, and the Projects page shows what it finds ("3 found", Clear search). ⌘K opens the command palette from the button in it.
• Studio, Projects and History come first; Create (Stories, Audiobook, Dub, Transcribe), Voices and Automate are sections that fold. The ⌘1–⌘9 keys are as before.
• No status bar: what goes on shows as cards at the foot of the sidebar — a stopped engine (Restart), dictation and calls under way (Stop, Hang up), each story, book, dub and workflow being made, takes, exports, transcriptions and downloads, several of a kind as "2/5", a finished one for 3 seconds. The pointer over them opens the stack; View all opens the new Tasks page, with every job that runs or waits.
• Under the cards: Settings, the logs panel (⇧⌘L) and Tools, and the notifications' bell (a dot while some are unread).
• Models and Integrations are Settings sections now (their links open them); Engine and Logs left the sidebar (the palette, the Go menu, Settings → Advanced → Diagnostics). ⌘E's quick switch is a small dialog with the engine's state and the model in use.
• Settings: 17 sections become 11, each a column of groups with a line under every setting: General (with Appearance), Studio (with Pronunciation), Dubbing, Dictation (with Permissions), Writing help, Integrations (apps and agents, then the Local API), Audio tools, Models & network (the models, then downloads, storage, the Hugging Face token, memory and licences), Privacy, Advanced (storage, usage, diagnostics and the self-check) and About. Old links open the part they were.
• Page titles are smaller; Projects and History have ‹ › (⌘[ ⌘]) and how many; Projects switches list and cards in its header (⌥1 ⌥2). Toasts show at the foot of the page, except on the Studio, Dub, Stories and Audiobook, whose foot has a bar of its own: there they show in the header.

Updates show in the window

• Settings → About → Software updates: Check for Updates…, the daily check, installing without asking (a version found by the daily check installs and waits for the restart) and beta updates as before, and Show Changelog…, every version's notes in AudioSmith's language. The choices made in the old update window carry over.
• Releases send their notes in English and Chinese (CHANGELOG.md and CHANGELOG.zh-Hans.md), so actnow.dev's changelog page and the app show each in its language.

Discover voices: about 35 chosen voices instead of 1,184

• Discover voices now lists a small set chosen by listening tests' measurements (thread #91's audit): Chinese and English voices first, a few in Japanese, Korean, Spanish, French, German, Italian, Portuguese and Russian. The 1,126 generated combinations and the weaker hand-made voices are gone from the list.
• Each designed voice has a written character (age, voice, situation) that the model really gets, a seed that was tried, and a reference that ships with the app: Use voice saves it ready to speak, with no reference to wait for. Preview plays a clip made from that reference, as the Studio would say it; nothing is rendered to listen.
• The gallery also offers the built-in speakers worth starting from (Serena and Dylan for Chinese, Kokoro's best English and European voices): Use voice saves the speaker and opens the Studio.
• Filters: use (narration and audiobooks, conversation and podcasts, characters and kids, ads and announcing), language, gender and age, and search. Pitch, accent or dialect, style and the nine categories are gone.
• Voices saved from the earlier gallery keep working (they carry their own recipe), and dub casts or agents that name an earlier gallery voice or design preset still get it: those ids are kept, just not listed.

Design: eight starting points

• Narrator, Conversational, Newsreader, Presenter, Storyteller, Elder, Villain and the new Soft (which replaces Whisper and Calm Guide). Upbeat, Kid and Northeast Buddy are gone (a Northeastern voice is in Discover voices). None asks for slow speech any more.

Fewer weak built-in speakers

• Kokoro's af_nicole (a whisper) and bm_lewis (rough) are no longer offered, and Qwen3-TTS's Ryan is not offered in Chinese (he gave up on Chinese numbers in tests). A project or a setting that uses one keeps it, and the API still takes them by name.

Settings breathe, and say it shorter

• Every Settings page follows one spacing: the groups' padding, room above and below each row with a line between rows, a group's note as a block of its own, and room between titles, lines and controls. Writing help's model picker no longer cuts its value, and the status line that repeated the rows is gone.
• 28 texts are shorter, in English and Chinese.

Designed voices no longer say leftover reference text

• Slow designed voices (Storyteller, Elder, Meditation guide, and a few gallery voices) were cut off half-way through their reference: Qwen3-TTS stopped at a length limit that slow speech reached (engine patch 0008 raises it from 0.48 s to 0.8 s per text token, and AudioSmith's own limit can now raise it as well as lower it). The reference was still saved with its whole text, so clones said the missing half before every line.
• A reference made from a voice must now say its text: when the recognizer hears none of the three takes say it (or they were cut off), they are made once more; a kept take that says something else gets what it says as its text, or none is kept and the voice says why. Its text is never one the audio doesn't say.
• References made before are heard once, the first time a take, story or book clones them: one that doesn't say its text is made again (a note says so) or gets the text it says.
• Design prompts no longer ask for slow speech ("speaks slowly", "语速偏慢") from a preset or a picked pace: it made references slow and clones sound read. A description you write keeps it.
• Voices designed for Japanese, Korean, German, French, Spanish, Italian, Portuguese or Russian say their reference in that language, not English.
• Gallery previews use the tuned sampling the Studio uses (temperature 0.7, top-p 0.8 for Qwen3-TTS) instead of the engine's defaults; previews are made again once.
• The Hindi gallery voice is hidden: no model here designs or clones Hindi.

Dictation's sounds to choose

• Settings → Dictation → Sounds: the sound of listening starting, of it stopping and of an error are each chosen from the ActNow family's sound library (Tally, Wrap and Shrug by default), macOS's system sounds or none. A compact picker searches them (names and tags), filters by type, plays each with ▶ (again to stop) and sets the volume in steps; Try plays the choice.
• The sounds are React Bits Sound Lab's and come with the builds from actnow.dev and GitHub releases only: they are not in this repository or under its licence (NOTICE). A build from source plays the system sounds AudioSmith used before (Tink, Pop, Basso).

Models move to the ActNow folder, out of Time Machine

• Models are kept in /Library/Application Support/ActNow/models, shared with the other ActNow apps, instead of the Hugging Face cache (/.cache/huggingface), which cleaning tools treat as a cache. Models already there are brought over at the first start without downloading them again (hard links: the cache stays as it is, for the tools that use it).
• Time Machine leaves models out (they can be downloaded again); Settings → Advanced → "Back up models with Time Machine" turns it back on for all ActNow apps.
• A model file cut short (a disk error, a copy that stopped) is no longer taken for a downloaded model; downloading again repairs it. A download of another version left half-way no longer makes a downloaded model look incomplete. Checking a model's files waits for other ActNow apps working on it and tells them when a file was bad.
• If the ActNow folder's model claims are deleted (a cleaning tool), AudioSmith claims its models again when it comes to the front, and keeps models another app may still use.

Audiobooks keep their rendered parts

• The parts of an audiobook already rendered are kept in AudioSmith's data folder instead of ~/Library/Caches, which cleaning tools empty: Resume after a stop or a failure no longer renders hours of a book again. Those of earlier renders move over at the first start.
• Temporary files of recordings and imports cut short are cleaned at start.

Keys stay out of logs

• Google Translate (Basic)'s key is sent in a header, not in the address, so a network error no longer writes it into the log. Logs and the diagnostic bundle leave out keys, tokens and proxy passwords.

Dub editor: ⌘Z and ⇧⌘Z undo and redo again

• Menu items with ⇧ in their keys show and take exactly those keys (⇧⌘Z is Redo, not also ⌘Z).

The local writing model answers in full

• The built-in writing model (Qwen3 4B) could cut its answers short or garble them when they had to follow a fixed form. The AI director left the last line of a story undirected; the emotion director could lose half a scene ("surprisedn”: 2note”: in place of an emotion); Cinematic translation's review copied the translation instead of pointing out problems, so nothing was improved. These came from how the engine kept answers to the form. They are fixed: in tests every line of every story is directed and the review finds real issues.
• The same request now gets the same answer every time; before, the order of an answer's parts changed each time AudioSmith started.
• An answer that starts repeating itself is stopped at once and asked again, instead of running for minutes (a glossary note did).
• A retry after an unusable answer is no longer a copy of the same answer: it is sampled.

The first public version, a beta: voice input for macOS on remote models, from actnow.dev. It updates to the next beta and to the release that follows (Settings → About → Beta updates). 0.1.0-beta.1 was built but not released; its changes are listed here.

Added

• Transcribe: a shortcut (fn by default; tap, double tap or long press, several per feature) records, a speech service recognises, a language model cleans up (fillers, false starts, punctuation, numbers; an optional structure layer and your own rules), and the text is pasted where the cursor is. Without a language model the text goes in as recognised; when it fails or is slow, too.
• Translate: the selection (read through Accessibility, or ⌘C where that is safe) or what you say, into the language you choose; a strict retry when the answer looks untranslated; an answer card to read, copy, insert or translate again into another language.
• Rewrite: say how to change the selection, or what to write with nothing selected; every answer is written one line under the pill, then waits in the answer card (as in Voxt), where Insert puts it in place (over the selection) and Continue records a follow-up, answered in plain text in the card's conversation.
• App enhancement: groups of rules by app and by website (address rules read through Accessibility, no browser automation), a scene prompt layered on top of the clean-up, the key to press after inserting, the features each group applies to, a live match preview, and the group's label in the overlay and the history.
• Dictionary: terms with aliases, sent as hints to the speech services and as a glossary to the language model; CSV and text import and export.
• Model services: Doubao and Alibaba Cloud Model Studio streaming recognition and any OpenAI-compatible transcription API; DeepSeek, Qwen, Volcengine Ark, Kimi, GLM, MiniMax, StepFun, MiMo, SiliconFlow, OpenAI, Anthropic, Gemini, OpenRouter and other OpenAI-compatible language models; keys in the Keychain, connection tests, a service per feature.
• Home: a slogan with your Transcribe shortcut in a key cap; what is still missing (a service, a permission, the shortcut), each line the way to fix it, and nothing once all is ready; a try of the whole chain on the page until the first result; today's results and characters, the last five results.
• Setup guide on the first start (and from Settings → General): privacy, permissions, services, then a real try of Transcribe, Translate and Rewrite, and App enhancement and the dictionary.
• History: by feature, searchable, with the texts, where the result went, the services, the rule that matched and each stage's time; recordings kept on request. Right-click an entry, in History or among Home's recent results, to copy its result or its recognised text, show its details, show its recording in Finder, retry it or delete it (with its recording, after a confirmation); a selected History row takes ⌘C and ⌫ too.
• The recording overlay (non-activating, so the app you type into keeps its focus), after Voxt's: the feature's icon and a waveform on a translucent, blurred pill with no buttons and Voxt's proportions (the shortcut stops, Escape cancels; the pointer goes through it), one line of live text under it, the pill as wide as the text up to 300 and the text then scrolled to its newest words, a loader while it works, errors in place of the waveform, nothing after an insert; and the answer card, with start and stop sounds. Every session starts as the pill, and it grows about its edge of the screen without moving; every flow shows the same pill until its result (what the model works on, then its draft, on one line), and the card comes only with the result.
• The answer card, after Voxt's: 440 wide (Voxt's is 560) and as high as its text (up to 220 of it, then it scrolls), its actions in the header (the target language as a capsule opening our own menu in a window of its own, Continue, Insert, Copy, Close), the header as far from the top as from the rule, Voxt's own icons in the pill and the card, why the result is in the card in Insert's tip instead of a line of orange, Copy marking itself for two seconds, Insert hiding the card while it pastes, another language translated in place (the old text with a spinner until the new one comes), and Rewrite follow-ups recorded in the card as a conversation (a small waveform, bubbles); Escape closes it, or cancels a follow-up only.
• Silence left out before recognition (a light energy detector, no model): a file service gets the recording without the silence around the speech and with long pauses shortened, a streaming service nothing before the speech starts, and a recording without speech is not sent at all (Settings → Audio → Skip silence).
• Sound optimization before recognition (pure DSP, no model): a high-pass filter at 80 Hz takes out hum and rumble, and one gain measured on the speech alone raises a quiet voice (up to +12 dB, never lifting the room's noise over -50 dBFS) or lowers a very loud one a little, with a limiter at -1 dBFS; a stream's gain follows slowly and holds through pauses. The recording kept in the history is the original; the history says what was done and the gain (Settings → Audio → Optimize sound, on by default).
• A small window (800×520 by default, 680×440 at the least) with no status bar: notifications stack as cards at the foot of the sidebar (a dot on Settings when it shows only icons), and a missing service or a stopped helper becomes a card with the action that fixes it.
• The menu bar icon: start or stop transcribing, paste the last result, the last three results to copy (one short line each, the whole in the tooltip); it changes while recording. Closing the window keeps Auralink there, and it can open at login.
• Settings: General (languages, the menu bar, login, the setup guide), Permissions, Audio (the microphones in your order of preference, skipping silence, optimizing the sound, sounds), Overlay style (place, opacity, corners, the recording time, with a live preview over a desktop picture), Shortcuts (every feature's keys and their conflicts), Appearance (the sidebar shrinks to icons in a narrow window), Network (proxy), Privacy, Storage, About (versions, updates, the self-check, reporting a problem); English and Simplified Chinese.
• Signed with a Developer ID and notarised. Updates from actnow.dev, in Auralink's own window: a daily check that only shows "Update available" at the foot of the sidebar (a dot on the icon rail), a panel with the new version's notes in Auralink's language (English where a version has no Chinese, marked so), a download of only what changed with its progress, installed when you say so, then Restart to Finish the Update; a failed download leaves Auralink unchanged and offers Try again. Check for Updates… in the app menu, the menu bar icon's menu, the palette and Settings → About, which also shows the changelog of every version in Auralink's language. An update installed while Auralink runs asks for a restart instead of starting the new helper.
• Beta updates (Settings → About): betas too, or releases only; on by default in a beta. Turning it off never goes back to an older version: a beta stays until a release newer than it. Install updates automatically (off by default) installs what the daily check finds and asks for the restart.
• Mute system audio while recording (Settings → Audio, on by default): the current output device is muted from just after the start sound until the recording stops, so that videos, music or calls are not recorded; a device you muted stays muted, one you unmute during the recording stays on, and a device that cannot be muted leaves the recording as it is (the overlay says so). No permission is needed. If Auralink or its helper stops while muting, the sound comes back at once, or at the next start.
• Sounds to choose (Settings → Audio → Play sounds): the start and the stop sound are each chosen from the ActNow family's sound library (from React Bits Sound Lab: taps, toggles, media, system sounds…; Tally and Wrap by default), or macOS's system sounds, or none. A compact picker searches them (names and tags), filters by type, plays each with ▶ (again to stop) and sets their volume in steps; a Try button plays the choice. A Voxt pair chosen in an earlier build is kept as those system sounds. The output is muted once the start sound has been heard, not after its file's silent end. The family's sounds are in the builds from actnow.dev; a build from the source code plays macOS's system sounds.
• History settings (the button beside History's search field): keep the recordings (the same switch as Settings → Privacy), choose the folder they are kept in (the recordings already kept move there, each one copied and checked before the original is deleted; if the folder is on a drive that is not connected, new recordings go to the default folder until it is back), see how much room they take, export them (all, or those of the tab shown) into a folder you choose, named by their date and first words, or delete them all and keep the texts. Automatic clean-up, on as in Voxt: entries older than 3 months are deleted with their recordings; you can choose 1 week to 1 year, keep at most 50 to 1000 entries of each feature, or turn it off. A new setting that would delete entries now asks first, with their number.
• Retry a history entry from its recording: a failed entry (or one with no text) shows Retry on its row instead of Failed, in its details and in its context menu; other entries with a recording have Transcribe again. The recording is recognised again (streaming services get it through a stream, faster than it was said), then cleaned up, translated or rewritten with your settings as they are now; the entry is updated in place (text, status, time), and nothing is pasted. The overlay's notice of a failed recognition opens the entry and retries it.
• When recognition fails, times out or finds no text, the entry keeps the recording for 7 days even when Keep the recordings is off, so that you can retry it; a successful retry deletes it.
• Logs (the Logs page and panel): click a line to copy it; Shift-click, ⌘-click or drag to select lines as in Finder, ⌘A for every line shown; right-click to copy a line, the selection or every line shown. The header shows how many lines are selected, with Copy and Deselect.

Changed

• The note of which output device Auralink muted is kept in Application Support instead of ~/Library/Caches, so that a cache cleaner run after a crash cannot leave your speakers muted. Logs leave out keys, tokens and proxy passwords.
• Auralink is free and open source, under the GNU AGPL-3.0: the source is on GitHub, and contributions are welcome (CONTRIBUTING.md; a contributor licence agreement). Versions up to 0.1.0-beta.1 were under the Apache License 2.0. Settings → About shows the copyright and the licence, links to the source code, the website and the changelog, and opens the licence and the third-party notices, which now ship inside the app. Report a bug opens a GitHub issue with the diagnostics filled in.
• Downloads, updates and release notes come from actnow.dev instead of GitHub Releases: sign in at actnow.dev/download/auralink to download. Updates come from actnow.dev/updates/auralink, and Beta updates read its beta channel, which always offers the newer of the latest beta and the latest release.
• The menu bar icon is now a speech bubble with a waveform, in thin lines like the system's own icons; while recording, the bubble turns solid.
• A result goes where you are typing when it is ready, not only into the app you started in (as in Voxt): move to another app and click into a text field while Auralink records or works, and it is pasted there; web editors such as Notion or Slack in the browser count as text fields. The app's group's key after the paste is pressed only in the app you started in, whose group shaped the result, and a selection is replaced only in the app it was selected in. Where no text field is focused (Finder, the desktop, a page you are reading) or the focus is in a password field, the result waits in the answer card, which says why; click where it should go, then Insert. Clicking Auralink's own window still sends it back to the app you started in. The history shows the app it was inserted into when that is another one. Insert in the answer card and Paste the last result follow the same rules.

Known limitations

• Apple silicon and macOS 15 or later only (no Intel build yet).

The apps share one copy of each model, some settings, and, if you agree, API keys, and they don't get in each other's way. AudioSmith names another app only when a released version of it is installed; with no other ActNow app, nothing on screen changes.

Downloads and updates move to actnow.dev

• AudioSmith now has its home on our website: actnow.dev/audiosmith. Download it from actnow.dev/download/audiosmith (free; the download needs a free account), and read every version's notes at actnow.dev/audiosmith/changelog. The source code stays on GitHub under the AGPL-3.0, and you can still build AudioSmith yourself.
• Updates come from actnow.dev: the daily check reads the version file there, and with Receive beta updates on also the betas' one, and offers the newer version. Copies of 0.4.0-beta.4 and earlier update to this version from GitHub, one last time.
• Settings → Diagnostics and about links to the website, the download page and the release notes.

Models shared with other ActNow apps

• On first run, if another app (say Dicta) has already downloaded some of the models AudioSmith needs, the setup says so ("3 models are already on this Mac · Installed by Dicta · no download needed") and uses them. Only the rest is downloaded. Choose models… lets you pick them yourself on the Models page.
• The Models page shows which models AudioSmith shares with other apps ("Shared with Dicta"), which ones another app has loaded right now ("Also loaded in Dicta"), and which ones another app installed ("Installed · from Dicta", with Use).
• Deleting a model that another app also uses asks what to do. Remove from AudioSmith keeps the files on this Mac for that app. Delete for all apps deletes them; the other app then has to download the model again. Clean up and "Erase everything" keep what other apps use, and say which app uses it. A model another app deleted for all apps shows "Deleted in Dicta · Download again".
• Settings → Storage shows how much space AudioSmith shares with other apps. It also offers to clean up models that only uninstalled apps used.
• If you accepted a model's licence in another app, AudioSmith still shows the licence, says when you accepted it, and asks you to continue.
• Kokoro's, Chatterbox's and Whisper's helper files now always load from the copy AudioSmith downloaded, at that version. Before, the engine could download them again on first use, or pick up another app's version.
• Downloads coordinate with the other apps: a file another app is downloading is not fetched twice ("Downloading in Dicta…"), and a model is not deleted while another app downloads or uses it.
• The engine no longer guesses which version of a model to load when several are on disk.

Settings shared with other ActNow apps

• The interface language, proxy, download source and model folder are shared with the other ActNow apps. With another app installed, each of these settings says whether it applies to all ActNow apps, and you can keep a different one for AudioSmith only.
• When another app keeps its models in a different folder, the Models page offers to move AudioSmith's models there. The move can be cancelled, and models the folder already has are not copied again.

API keys shared with other ActNow apps

• API keys and the Hugging Face token can be shared with the other ActNow apps. With another app installed, a checkbox under the key ("Let other ActNow apps on this Mac use this key") chooses whether to share it.
• AudioSmith offers a key saved in another app ("Use the DeepSeek key from Auralink"). It reads that key only after you click Use this key.
• Removing a key that another app also uses asks whether to stop using it here or to delete it for all apps. If another app deletes a shared key, AudioSmith asks you for a key again. "Erase everything" never deletes another app's key.
• The Hugging Face token is looked up in this order: $HF_TOKEN, the shared token, AudioSmith's own, the huggingface-cli token file.

Shortcuts and memory

• If another ActNow app uses the same keys as the dictation shortcut, Settings → Dictation says so. The app that took the keys last shows the notice once, with Change shortcut, Open Auralink settings and Ignore. While you record a new shortcut, "Also used by Auralink (Transcribe)" warns you, but you can still save it.
• The ActNow apps share one memory budget. Before AudioSmith loads a model, it unloads its own idle models first. If that is not enough, it asks the other apps to unload models that have been idle for a minute. It does the same when they ask, and never unloads a model a job is using.
• Dictation's paste no longer overwrites the clipboard while another ActNow app is pasting.

Dub: conversations recorded together (multi-speaker dialogue)

• A new switch in the Dub page's Cast step, 「多人对话配音」 (Multi-speaker dialogue): the people of a conversation are recorded together by the dialogue model "Dialogue · Light" (SoulX-Podcast), each line hearing the one before it, so answers sound like answers. Each line is then fitted to its own time as before (Smart fit and the other timings) and placed where it was said, so the video's pauses and overlaps stay. Monologues, one person speaking alone and lines over 4 s apart stay line by line.
• On by default for new dubs into Chinese or English with two speakers or more; dubs made before keep their audio (off). Each speaker is cloned from about 12 s of their own lines, with what they say there, even from another language (an English or Japanese video dubbed into Chinese); a saved or designed voice from its own reference.
• A line the dialogue model does not say clearly is made again (after the lines before it, another seed), then line by line; such a line is checked too, and marked "Listen" when it is not clear either. Editing a line records it alone again after the lines before it; changing a voice records the conversation again. The line list shows each conversation's lines with a coloured bar and a 「对话 N」 chip; each line's ⋯ menu can record the whole conversation again or keep a line out of it.
• Greyed with the reason on Macs under 16 GB, with the 「逐句」 voice match, and for other languages (a Japanese track stays line by line; a dub into Chinese and Japanese uses it for Chinese only). Not downloaded yet: the switch offers the download (2.4 GB), and dubs are made line by line until it is there. The Generate step's model picker lists it too. Slower than line by line.
• The translation of such a track aims a little shorter: the dialogue model speaks about 1.01–1.12× the estimate (learnt from the lines it makes).
• Local API and MCP: create_dub, edit_dub and batch_dub take dialogue (auto or off) and dialogue_model (soulx), edit_dub a line's no_dialogue; get_dub returns dialogue, dialogue_blocked, dialogue_stretches and each line's dialogue, made_by (dialogue or single), fallback_reason and flag; run_dub with segments naming a conversation line records it again after the lines before it.

The first signed release is a beta (a GitHub prerelease): copies that receive beta updates get the next beta, and 0.4.0 once it is out.

Updates

• AudioSmith updates itself: once a day it reads the version file of its latest release on GitHub and, when a new version is out, shows its notes (this file's section of the version) and offers to install it, skip it or remind you later. Updates are signed; AudioSmith checks the signature before it installs one, and usually downloads only what changed. AudioSmith → Check for Updates…, the command palette and Settings → Diagnostics and about check at once; that page also turns the daily check and installing without asking on or off. Copies built on your own Mac don't update themselves.
• Releases are signed with a Developer ID and notarized by Apple, so they open without Gatekeeper's warning (docs/release.md).
• Beta updates: Settings → Diagnostics and about → Receive beta updates. Betas get new features sooner but may be less stable; turned off, AudioSmith only updates to releases. It is on in a beta copy (such as this one) and off in a release until you choose. Turning it off never goes back to an older version: a beta stays until a release newer than it is out. About shows a Beta badge next to a beta's version.

Transcription with speakers

• New model "Transcription with speakers" (区分说话人转写; MOSS-Transcribe-Diarize 0.9B, 8-bit, 1.3 GB, Apache-2.0): it transcribes and tells who says what in one pass. Download it from the Models page, Transcribe or the Dub page; nothing downloads by itself.
• Transcribe: a Speakers option in the toolbar. Transcripts list their speakers (S1, S2…) with their lines; click one to name it. Copy and TXT export put one paragraph per turn after the speaker's name, SRT cues start with it, VTT marks voices, JSON has each segment's speaker. Dictation and plain transcription stay on Qwen3-ASR (MOSS is about 4× slower: a 10-minute recording takes about 4 minutes on a MacBook Air M4).
• Dubbing: with the model downloaded, a source with more than one speaker is transcribed by it and every line gets its speaker from it (in our tests 1.00 of the lines right on three of four 2–4 speaker dialogues and 0.93 on a 13-minute one with five people, where Qwen3-ASR + Sortformer managed 0.50–0.69 and never found more than 2 speakers; no limit of 4 speakers); the aligner still times the words, and the glossary's terms are its vocabulary. Without it, the dub works as before (Qwen3-ASR + Sortformer) and suggests the download, then Transcribe again.
• Long recordings go in 10-minute windows that overlap by a minute, linked by who speaks in the overlap; stretches without speech are skipped, and the model's chat answers to silence ("I'm sorry, I can't assist…") are dropped.
• Local API and MCP: speakers=true, model=gpt-4o-transcribe-diarize or response_format=diarized_json; segments carry speaker. MCP transcribe takes speakers: true.

Fewer transcription and writing models; Qwen3-ASR transcribes everything

• Speech recognition is Qwen3-ASR 0.6B ("Chinese & English transcription") everywhere: dictation, Transcribe's Fast and Accurate modes, dubbing and its read-back check, workflows, Convert, clone references and calls. It made 0.5–4 % errors on Chinese where Whisper made 8–33 %, punctuates, and is 2–4× faster (thread #44's measurements). Whisper large-v3 turbo ("Multilingual transcription", tagged Fallback) is used on its own for the languages Qwen3-ASR does not know (it knows 30) and for translating speech into English. Fast and Accurate now differ only in the word timings.
• Long files: the vocabulary is only sent with recordings up to 30 s (Qwen3-ASR copied it into the end of longer transcripts); the " The." (or "嗯。") Qwen3-ASR sometimes invents after the last sentence of a long file is dropped; files longer than about 4½ minutes are timed word by word in windows of at most 4 minutes, cut between sentences (the aligner's times broke down after about 4½ minutes).
• Removed: Parakeet TDT 0.6B v3 (no Chinese) → Qwen3-ASR, or Whisper for a language Qwen3-ASR does not know; Mel-RoFormer (its output ended in 2–3.5 s of silence, and it was 5–9× slower) → HTDemucs; Gemma 4 E4B, Gemma 4 E2B and Qwen3 1.7B (slower or broken JSON, weaker translations and clean-up) → Qwen3 4B Instruct 2507, now the only language model, named in Settings → Writing help instead of a choice. Settings that named a removed model go back to the default once, with a notice; a model you chose that is still there (Whisper) stays. API and MCP: a transcription that names Parakeet is answered by its replacement with replaced_by and a warning (X-AudioSmith-Replaced-By / X-AudioSmith-Warning); GET /api/models/{id} describes the removed models, and they are not listed. Their files stay until Models → Clean up, which now also lists them (and shows on their tab).
• Onboarding: Balanced no longer downloads Whisper; Max adds it instead of Mel-RoFormer.

Models named for what they do

• Every model has a plain name that says what it is for — Standard voices, High-fidelity clone, Fast clone, Voice design, Fast English voices, English clone, Multilingual all-rounder; Chinese & English transcription, Writing assistant · Standard… (标准音色、高保真克隆、快速克隆…) — a one-line description and a few tags (Chinese, Cloning, Built-in voices, Low memory, Non-commercial…). Pickers, the Models page, ⌘E, Settings, the bar beside Generate, History, onboarding, notifications and messages use them.
• The family, version, size, quantisation, repository and licence are still there: in a model card's Details on the Models page, in the pickers' tooltip and ⓘ, and in History's details (Technical name). Ids, repos and saved projects don't change.
• The model picker and the Models page have a search field that finds a model by its name, its technical name, id, repo or a tag.
• Local API and MCP: model lists add display_name (English), ui_name (as the app shows it), description and tags; a friendly name picks a model too.

Portable voices, and simpler Design, Clone and Export

• Make a portable copy of a model's built-in speaker or of a designed voice (My voices → ⋯, the Studio's voice list, or Make portable where a voice can't be used): its model says a ten-second line, evened out, and the take is saved as a cloned voice "<name> (approximate)" that every model that clones can speak with. It sounds close to the original, not the same, and says so.
• When a voice can't be used with a model, the message says why and what to do (make a portable copy, or pick a model that designs).
• Design shows only what the model can't express; what it will hear is behind ⓘ, and the seed moved into Details. Clone: style and language are under Advanced. Export audio: format, loudness, file name and extras in view; bit depth or bitrate, sample rate, channels and trimming under Advanced.

Simpler Stories, Audiobook and Dub

• Stories and Audiobook: the inspector keeps Setup, Cast, Performance (Stories) or Chapters and Cover & details (Audiobook), and one Advanced section for the rest — language, merging and evening out voices, refinement and guidance, gaps, trimming, seed, a stem per character, a story's M4B with its cover and details, an audiobook's own pronunciations — with N changes and Restore recommended (your pacing stays).
• The Project section moved to a Project ▾ menu in the header: Save (under the title — the title is the project's name), Save a copy, New, recent projects, All projects…
• A story shows only what its model hears: personas, the scene, the AI director and line directions disappear on models that say lines as written, and a line's speed on models without one.
• Dub: each step keeps its main choices; the rest folds into Advanced at the foot of the step (speakers and review, more languages and the translator's options, voice match, the voice model's parameters, a WAV's bit depth). The buttons the editor's ⋯ menu already had left the panels; Transcribe again… and Translate everything again joined that menu.

A simpler Studio

• The inspector shows what most takes need: the model, one Voice list (the model's speakers, your saved voices, a clip to clone, Manage voices…) and the speed. Everything else — style, reference transcript, length, refinement, guidance, sampling, seed, reading numbers, evening out the volume — is under Advanced, filled with the model's recommended settings; its header says "Recommended settings" or how many you changed, with Restore recommended inside. Advanced stays open or closed as you left it.
• The Voice controls and Audio quality buttons are gone, and so are the WAV precision (takes are kept at 24-bit; choose the file's bit depth in Export audio), Preload / Unload (in the ⌘E engine switcher, now with Unload), the autoplay switch (Settings → Studio) and the steps slider (three tiers remain).
• A clip chosen to clone is transcribed for you when it is a WAV.
• The bar beside Generate says which model and voice will speak, and how many settings you changed; a click shows the inspector.

Recommended models and settings

• Every speech model was measured on a MacBook Air (intelligibility with two recognisers, naturalness, voice stability, speed, memory). Each page now recommends two or three models for its job and your language, with a one-line "best for" and a speed and quality hint; the others are one click away under "Show all models", and the ones that can't do the job (no cloning for a dub) are greyed out with the reason.
• Tuned defaults: Qwen3-TTS uses temperature 0.7 and top-p 0.8 unless you (or an API request) set them — the models' own 0.9 / 1.0 sounded worse and less stable; OmniVoice uses guidance 1.5 with 32 steps (16 for dubbing). The Local API and MCP tools follow the same defaults. Rendered chunks from before are made again once with the new settings.
• Removed: six speech models that misread, paused or covered too little in the tests — KittenTTS mini, MOSS-TTS-Nano, Chatterbox (multilingual), Marvis, Soprano and Pocket TTS. Seven remain: Qwen3-TTS 0.6B CustomVoice, 0.6B Base, 1.7B Base and 1.7B VoiceDesign, Kokoro, Chatterbox Turbo and OmniVoice (under Show all models). Drafts, projects, dubs, workflows, settings, history and API or MCP calls that name a removed model use its replacement — Kokoro, Qwen3-TTS 0.6B Base for MOSS-TTS-Nano, Chatterbox Turbo for Chatterbox and for cloning with Marvis, Soprano or Pocket TTS — and the app says so once; the API answers with replaced_by and a warning. A removed model's own speakers saved in My voices can't be used any more.
• Models → "Removed models still take X GB": Clean up deletes their files (after listing them); nothing is deleted on its own.
• Chatterbox's expressiveness control is gone with the multilingual model (Turbo does not take it).
• New dubs fit lines with "Smart fit" instead of "Lip sync": lip sync sped most Chinese lines up audibly (1.2–1.4×). Dubs you already made keep their timing.
• Kokoro's six Chinese voices are hidden (unintelligible in the tests); its English voices are the best English default.

Stories: dialogue that flows

• Turn-aware pacing: the silence between lines follows the conversation — shorter when a character goes on or answers a question, a little longer after a statement, shortest when a line is cut off (… —), longer around narration. Three presets (Dialogue, Reading, Relaxed) in the new Performance section; new stories use Dialogue, stories made before keep their fixed gap until you pick a preset. A line's own "pause after" still wins.
• Consecutive lines of one character are said in one go, so the voice flows from one sentence into the next, then split back into lines at their pauses (player, stems and the sidecar see lines as before; a turn that cannot be split is made line by line). The pause inside a turn is kept close to the preset's.
• Each character's lines are brought toward one loudness (measured 4–7 LU apart on Qwen3-TTS before, about 1 after).
• A scene for the story, a persona for each character and a direction (with an intensity) for each line. Qwen3-TTS CustomVoice and VoiceDesign hear them as an instruction with every line; other models say lines as written, and the page says so ("Line by line"). Never spoken.
• AI director: the writing-help language model (Qwen3 4B on this Mac by default) reads the story and fills in the scene, personas and directions; edit any of them, Undo, or direct again.
• Imported scripts keep stage directions out of the spoken text: 妈妈（温柔地）：…, MOM (softly): …, [Mia] (pleading) … become the line's direction.
• Agents: update_longform sets a story's scene, personas, directions, intensities and pacing, and direct: fill | all runs the AI director; get_longform shows them with the model's continuity level.
• The catalogue marks Qwen3-TTS CustomVoice and VoiceDesign as taking instructions, as the engine reports them; dubs and stories ask one predicate.

Export audio

• History's takes have an Export button (and Export… in ⋯); the Studio's latest take, a take's details and Projects open the same Export audio dialog, which replaces the Save as… menu.
• Choose the format (WAV; M4A through the engine; MP3, FLAC and Opus with ffmpeg, greyed out with a pointer to Settings → Audio tools without it), the bit depth or bitrate, the sample rate and the channels; bring the file to −16, −18 or −23 LUFS with a true-peak ceiling; trim the silence at both ends; name the file and its title and artist tags; save the script and subtitles beside it. The dialog shows the size and length before you export, and remembers your choices for each kind of take (Studio, voice preview, converted recording, Local API).

Writing help no longer fails during a render

• The language model on this Mac now waits for its turn when a render, dub or batch is using the engine, instead of failing with "the engine is busy": writing help, the AI director, dub translation, transcript clean-up, the direction coach, phone calls and API/MCP requests go next, ahead of queued batch work (a running job is never interrupted). These short requests don't appear in the jobs list. Their time limits count from when the model starts.
• Dictation never waits for a render: while one runs, it types the polished text at once without the model's clean-up. The Tools direction coach shows 「等待当前任务完成…」 (Waiting for the current job to finish…) and, if the job is still running after 30 s, reads the note by keywords.