tautau docs
Everything you need to dictate into any text field on the web. Install the extension, click the mic, speak. The sections below cover setup, voice commands, snippets, your dictionary, history, quotas, and what to do when something misbehaves.
Getting started
- Install tautau from the Chrome Web Store, then pin the extension so it stays visible in your toolbar.
- Open any page with a text field. A mic button appears next to the field when you focus or hover it. Click it to start recording.
- Prefer the keyboard? Press Cmd+Shift+V on macOS or Alt+Shift+V on Windows. You can also right-click an editable field and choose "Start voice dictation".
- Allow microphone access when Chrome asks.
- While you record, a live level meter shows your input. Stop and you get instant playback of what you said, then the transcript lands in the field.
Choosing a microphone
The extension options page lists every input device Chrome can see. Pick the one you want and hit the test button: a level meter shows you the signal before you dictate for real. If you plug in a new mic, reopen the options page and select it there.
AI polish
After transcription, an optional cleanup pass fixes punctuation and capitalisation and removes filler words. It never changes what you meant. If the cleanup step fails for any reason, you simply get the raw transcript instead.
Polish is on by default. Toggle it in the extension options, or with the Polish switch on the tautau.xyz playground. Polished transcripts carry a "polished" badge in the playground session log so you can tell which pass ran.
Voice commands
Speak punctuation and editing commands as you dictate. Commands are case-insensitive and work mid-sentence, so you can say "thanks comma talk soon" and get exactly that.
| You say | You get |
|---|---|
| period / full stop | . |
| comma | , |
| question mark | ? |
| exclamation mark / exclamation point | ! |
| colon | : |
| semicolon | ; |
| open quote / close quote | " |
| new line | Moves to a new line |
| new paragraph | Starts a new paragraph |
| scratch that / delete that | Removes the last sentence |
| delete word | Removes the last word |
You can turn voice commands off in the extension options if you regularly need to say words like "period" literally.
Voice snippets
Say "insert" followed by a trigger word and tautau expands it into saved text. Manage your triggers and expansions in the extension options. Some ideas to start with:
- "insert signoff" for your email sign-off
- "insert address" for your postal address
- "insert standup" for your daily standup template
- "insert reply" for a canned support reply
Personal dictionary
Sign in and open tautau.xyz/settings to teach tautau your vocabulary. Add names, jargon, and product terms, up to 200 of them. Your terms are used as preferred spellings during AI polish and as a recognition hint during transcription, so "Ebsworth" stops coming out as "Epworth".
History
Signed-in dictations land in tautau.xyz/history: a searchable list of past transcripts with audio replay. Delete entries individually or clear everything at once. History is for signed-in users only. Anonymous dictations are never stored, full stop. Read the privacy policy for the details.
Transcribing a URL
Dictation is not the only way in. tautau.xyz/transcribe turns a link into a transcript: paste a YouTube video, a podcast episode, an RSS feed, or a direct audio file, and you get the full text with Copy and Download buttons.
- YouTube captions: fetched directly from the caption track, so the transcript comes back in a second or two. Free to try: 5 transcripts with no account, 5 a day with a free account, 20 a day on paid plans.
- Podcasts and audio files: transcribed by AI, which runs async. The page shows a progress state and polls until the transcript is ready; long episodes can take a few minutes.
AI transcription is metered in minutes because it costs real compute: free accounts get 15 minutes a month, Base and Pro use the same 200 or 600 minute allowance as dictation. Anonymous visitors can use the captions path only; the page asks you to sign in when a link needs the AI path. The result card always shows a badge ("captions" or "AI transcription") so you know which engine produced the text.
MCP server for AI agents
tautau ships an MCP (Model Context Protocol) server so AI assistants like Claude Code, Claude Desktop, and Codex can transcribe links and audio files directly inside a chat. Ask your agent to "transcribe this YouTube video" and the full text lands in the conversation.
Install in thirty seconds with the setup script, or add it by hand:
curl -fsSL https://tautau.xyz/mcp/install.sh | bash # or by hand, per client: claude mcp add tautau -- npx -y tautau-mcp # Claude Code # Codex, in ~/.codex/config.toml: # [mcp_servers.tautau] # command = "npx" # args = ["-y", "tautau-mcp"]
- transcribe_url: paste a YouTube, podcast, RSS, or direct-audio link, get the full transcript in the chat.
- transcribe_audio: point at a local audio file (mp3, wav, m4a, ogg, webm) and get it transcribed.
- get_quota: check how much of the free allowance is left.
The server talks to tautau.xyz anonymously, so the usual free quotas apply (5 dictations and 5 YouTube transcripts in total). When you hit the wall it tells you how to continue with a free account. Source and full reference: npmjs.com/package/tautau-mcp.
Accounts and quotas
- Anonymous: 5 dictations and 5 YouTube transcripts in total, no account needed.
- Free account: 5 dictations and 5 YouTube transcripts a day, plus 15 podcast/audio minutes a month. No credit card required.
- Base ($3/month): 200 audio minutes per month.
- Pro ($9/month): 600 audio minutes per month.
Hit a limit and the error takes you straight to /pricing to upgrade. Prefer your own provider? BYOK (bring your own key) bypasses tautau quotas entirely: configure your STT API key in the extension options.
Troubleshooting
The level meter flatlines at exactly zero
You have hit a known Chrome/macOS audio-pipeline bug where the mic opens fine but captures pure silence. Quit Chrome fully (Cmd+Q) and reopen it. Unplugging and replugging the mic also clears it.
No mic button appears
Refresh the page after installing the extension, and check the extension is enabled in your Chrome toolbar. The button attaches when you focus or hover a text field.
Microphone permission denied
Check Chrome's site settings for microphone access and allow the site you are dictating on, then try again.
Transcription fails with a quota error
You have used up your plan's allowance. Head to /pricing to upgrade, or set up BYOK to bypass quotas.
FAQ
Is my audio stored?
Only if you are signed in. Signed-in dictations are saved to your private history with audio replay, and you can delete them individually or clear everything from tautau.xyz/history. Anonymous dictations are never stored: the audio is transcribed and discarded.
Does tautau work offline?
No. Transcription runs on the server, so you need an internet connection while you dictate.
Which browsers are supported?
tautau ships as a Chrome extension and works in Chrome and other Chromium-based browsers that support extensions from the Chrome Web Store.
What does it cost?
You can dictate anonymously for 5 dictations in total, or create a free account for 5 dictations a day with no credit card. Base is $3 per month for 200 audio minutes, and Pro is $9 per month for 600 audio minutes. See tautau.xyz/pricing.
Can I use my own speech-to-text API key?
Yes. Bring-your-own-key (BYOK) lets you plug in your own STT provider key and bypass tautau quotas entirely. You configure it in the extension options.
Want deeper reading? The blog covers voice typing technique, accessibility, and productivity.