Facts
Dictation in your apps: facts and numbers
372 facts, each traced to the exact words of its source.
- A 2023 Chrome Unboxed guide to Chromebook Dictation lists the voice command "New line", which moves the cursor to a new line.Michael Perrigo, Chrome Unboxed, 2023
- A 2023 Chrome Unboxed guide to Chromebook Dictation says that, besides words, you can speak punctuation to Dictation.Michael Perrigo, Chrome Unboxed, 2023
- A GitHub report from March 2026 says voice dictation in Claude Code version 2.1.83 activated but never transcribed, and that rolling back to 2.1.81 fixed it for that user.GitHub, anthropics/claude-code issue tracker, 2026
- 60 megabytesA January 2025 reply in the Chromium thread estimates each on-device speech recognition language pack in Chrome on Windows, Mac and Linux at about 60 MB.Chromium blink-dev mailing list, 2025
- 300 minutes per monthA Microsoft Community moderator said in November 2023 that there is still a 300-minute monthly limit for uploading dictation files in Word, and that developers had not yet removed it.Microsoft Community moderator, Microsoft Q&A (Microsoft Learn), 2023
- 2021 year of the replyA Microsoft Q&A reply posted in September 2021 says that the Dictate function is available in the free online version of Word.Microsoft Q&A, 2021
- A Mozilla wiki page revised in January 2022 says Chrome, Edge, Safari and Opera supported the speech-to-text form of the Web Speech API, but Firefox did not.Mozilla Wiki, 2022
- A report quoting Talon's developer says Talon will soon remove all Linux support from its public release.OSNews, 2026
- A September 2026 Ubuntu how-to guide says Myna is an opt-in feature in Ubuntu 26.10, and that the current Ubuntu 26.04 can also try it out.UbuntuHandbook, 2026
- 5 times as longA Slack clip can run five times as long as a Teams clip: five minutes against one.Calculated from other facts
- 2 weeksA software engineer with severe hand pain wrote in 2025 that they found success with voice coding in two weeks, using four tools.whitep4nth3r (personal blog), 2025
- A user's bug report, opened June 4, 2026 and closed as not planned on July 10, 2026, says Claude Desktop dictation on macOS never produced a transcription.claudeissues.com (mirror of Claude Code issues), 2026
- A user's feature request about the Claude Code desktop app, opened in June 2026, says built-in dictation appears to be English-only, or auto-detects with no way to override the language.GitHub, anthropics/claude-code issue tracker, 2026
- A W3C TAG review request for on-device support in the Web Speech API was opened on 9 January 2025.W3C Technical Architecture Group (review request), 2025
- A Windows user's bug report, opened August 30, 2026 and closed as not planned on September 2, 2026, says Claude Desktop dictation worked only on the first attempt after launching the app.claudeissues.com (mirror of Claude Code issues), 2026
- According to the article, all of Myna's process works offline.UbuntuHandbook, 2026
- After a model is downloaded, Vocalinux runs speech to text on your machine with whisper.cpp by default, and it also supports OpenAI Whisper, Faster Whisper, VOSK and Parakeet.GitHub, VocaHQ/vocalinux, 2026
- After installation, Serenade walks you through installing plugins for supported applications such as VS Code, Chrome and Hyper.Serenade, n.d.
- 50 megabytes per language modelAlpha Cephei says Vosk's per-language models are about 50 MB each, and that much bigger server models are also available.Alpha Cephei, 2026
- Andrej Karpathy's original post described vibe coding as a new kind of coding where you fully give in to the vibes, embrace exponentials, and forget that the code even exists.Simon Willison, Simon Willison's Weblog, 2025
- Anthropic says dictation in the Claude mobile apps is available to all Claude users on the Free, Pro, Max, Team and Enterprise plans.Anthropic, Claude Help Center, 2026
- Anthropic says that in the mobile apps it deletes the audio recording after converting speech to text, and does not keep a copy of the speech or use it for training.Anthropic, Claude Help Center, 2026
- Anthropic's Claude Academy tutorial describes dictation in the Claude desktop app as pressing the mic, speaking, and then sending the message.Anthropic, Anthropic (Claude Academy), 2026
- Anthropic's Claude Academy tutorial says Claude dictation is available on the web, on mobile and in the desktop app.Anthropic, Anthropic (Claude Academy), 2026
- Anthropic's Claude Academy tutorial says that on a phone you hold the mic while you talk and let go to send.Anthropic, Anthropic (Claude Academy), 2026
- Anthropic's voice mode help article says voice mode supports many more languages, on all plans and on Claude Mobile, Claude Desktop and the web.Anthropic, Anthropic, 2026
- Apple says that on a Mac you place the insertion point in an app where you want to enter text, then start Dictation.Apple Support, n.d.
- Apple says that when a third-party app tries to access and control a Mac through accessibility features, the user must specifically grant that app access in Privacy & Security settings.Apple Support, Mac User Guide, 2026
- Apple's developer note says a process should never enable secure event input for general text input, and should turn it on only when it is needed.Apple Developer Documentation, Apple Developer Documentation Archive, n.d.
- 20 secondsAt 153 words per minute, 52 words take about 20 seconds to say.Calculated from other facts
- At a terminal prompt, VS Code's built-in dictation starts with the command Voice: Start Dictation in Terminal, run from the Command Palette.Visual Studio Code documentation (Microsoft), 2026
- By default, a browser may choose whether speech recognition runs on the device or on a remote service, because processLocally is false unless a page sets it to true.MDN Web Docs (Mozilla), 2026
- By default, speech recognition on a web page sends your audio to a server-based recognition service, so it does not work offline.MDN contributors, MDN Web Docs (Mozilla), 2026
- Canonical, the company behind Ubuntu, is developing Myna, an app for voice typing.UbuntuHandbook, 2026
- Canonical's roadmap for Ubuntu 26.04 LTS says Wayland is now the default session across modern GPUs.Canonical, Ubuntu Discourse, 2026
- 113 Chrome versionChrome 113 made WebGPU available on ChromeOS devices with Vulkan, Windows devices with Direct3D 12 and macOS, and Google said Linux and Android would come later.Chrome team, Chrome for Developers, 2023
- Chrome says content scripts are files that run in the context of web pages, and can read and change the page's content.Chrome for Developers, 2026
- Chrome's 139 release notes say the Web Speech API gained on-device speech recognition, so a website can make sure neither audio nor transcribed speech is sent to a third-party service.Google Chrome for Developers, 2025
- Chrome's developer guide says the activeTab permission does not display a permission warning, because it grants only temporary access to the site the user is on.Google Chrome for Developers, 2024
- ChromeOS Dictation can only be used when keyboard focus is on an editable field, such as a text area or an input box.Chromium Project (Google), undated
- ChromeOS Dictation turns itself off when no speech has been recognized for a short period. The document does not give the length of that period.Chromium Project (Google), undated
- Chromium's developer documentation describes ChromeOS Dictation as an accessibility feature that lets users type and edit text with their voice.Chromium Project (Google), undated
- 2 ways to turn Dictation onChromium's developer documentation says ChromeOS Dictation can be turned on in two ways: by pressing Search + D, or by pressing the microphone icon in the status tray.Chromium Project (Google), undated
- Chromium's developer documentation says Pumpkin is a semantic parser that lets ChromeOS Dictation extract meaning from recognized text.Chromium Project (Google), undated
- Chromium's developer documentation says that when Pumpkin is not available or fails to download, ChromeOS Dictation falls back to regex-based speech parsing.Chromium Project (Google), undated
- Chromium's developer documentation says the SODA component of ChromeOS Dictation turns speech into text on the device, without sending it to a Google server.Chromium Project (Google), undated
- Chromium's Dictation source tracks, for each locale, whether offline speech recognition is supported and whether the matching language pack is installed.Chromium Project (Google), undated
- Chromium's source code comments that the languages in its ChromeOS Dictation speech list are, by default, not supported offline. Offline languages are added only when the on-device speech installer reports them as available.Chromium Project (Google), undated
- Claude Code adds the current project name and git branch name to the recognition as hints, without you doing anything.Anthropic, Anthropic, 2026
- 10 secondsClaude Code pauses voice dictation after three failures within 10 seconds, and the pause lasts until 10 seconds have passed since the first failure.Anthropic, Anthropic, 2026
- Claude Code voice dictation needs a local microphone and does not work in cloud sessions or SSH sessions.Anthropic, Anthropic, 2026
- Claude Code voice transcription does not use Claude messages or tokens, and does not count toward the limits shown in the usage view.Anthropic, Anthropic, 2026
- Claude Code's dictation uses the same language setting as Claude's response language, and defaults to English if that setting is empty.Anthropic, Anthropic, 2026
- Claude Code's docs say /voice is available only while a claude.ai account is your active sign-in, so an unknown /voice command usually means you are not signed in with one.Anthropic, Anthropic, 2026
- Claude Code's docs say hold-to-talk depends on key-repeat events from the terminal, and recommend switching to tap mode with /voice tap when that is not available.Anthropic, Anthropic, 2026
- Claude Code's hold-to-talk mode works by detecting key-repeat events from the terminal, so it needs the terminal to send them.Anthropic, Anthropic, 2026
- Claude Code's speech-to-text service is available only when you sign in with a Claude.ai account, not with an API key or a cloud provider.Anthropic, Anthropic, 2026
- Claude Code's voice dictation streams the recorded audio to Anthropic's servers for transcription, and the audio is not processed on the computer.Anthropic, Anthropic, 2026
- 20 languagesClaude Code's voice dictation supports 20 languages, listed in its documentation from Czech to Ukrainian.Anthropic, Anthropic, 2026
- Claude Code's voice transcription is tuned for coding vocabulary in both hold and tap modes.Anthropic, Anthropic, 2026
- Claude's current API models accept text and image input and produce text output, and the models overview lists no audio input.Anthropic, Anthropic (Claude API documentation), 2026
- Claude's help center says that in Claude Cowork and Claude Code, dictation is available but voice mode is not.Anthropic, Anthropic, 2026
- 12 languagesClaude's mobile help article lists 12 languages for dictation on iOS and Android.Anthropic, Claude Help Center, 2026
- Claude's mobile help article says support for dictation in languages other than English is in beta.Anthropic, Claude Help Center, 2026
- Claude's mobile help article says to start dictation by tapping the microphone icon on the right side of the chat input.Anthropic, Claude Help Center, 2026
- Cursor's 2.0 changelog says Voice Mode lets you control the Agent with your voice, using built-in speech-to-text conversion.Cursor, 2025
- Cursor's Voice Mode lets you define custom submit keywords in settings, which trigger the Agent to start running.Cursor, 2025
- Cursorless describes itself as a spoken language for structural code editing that lets developers code by voice.Cursorless contributors, GitHub, 2026
- Dictate in Word marks some words that it may have misheard, and you can select Ignore on the marking if the word is already correct.Microsoft, 2026
- Every Cursorless command consists of an action performed on a target.Cursorless documentation, n.d.
- For Dictate in Word, Microsoft's accuracy checklist says to use a fast and reliable internet connection, reduce background noise, speak more deliberately and check whether the microphone needs an upgrade when many words are wrong or missing.Microsoft, 2026
- Google says Live Caption in Chrome works offline, including for audio and video files saved on the computer.Google Keyword blog, 2021
- Google says some Chromebook Dictation voice commands are not yet supported in right-to-left languages.Google (Chromebook Help), 2026
- Google says that to run a Chromebook Dictation command, you wait for your spoken text to clear from the box, then say the command.Google (Chromebook Help), 2026
- Google says the captions Live Caption creates in Chrome are generated on the device, so the audio does not have to leave the computer.Google Keyword blog, 2021
- Google says the newer Chromebook Dictation editing commands, such as Delete all and Move to the start, work only in English, French, German, Italian and Spanish.Google (Chromebook Help), 2026
- Google says you can add common punctuation to Chromebook Dictation by saying "comma," "period," "question mark," "exclamation mark" or "exclamation point."Google (Chromebook Help), 2026
- Google's Chrome help page says Chrome is compatible with devices that run Windows and Mac operating systems.Google, 2026
- Google's current Docs help page lists Spanish as many separate entries by country or region, such as Spanish (Argentina), Spanish (Bolivia) and Spanish (Chile), in its list of languages and accents for voice typing.Google (Docs Editors Help), 2026
- Google's current Docs help page says that if voice typing is not enabled in your organization, an administrator may have turned it off.Google (Docs Editors Help), 2026
- Google's current Docs help page says your computer microphone must be on and working before you use voice typing or voice commands.Google (Docs Editors Help), 2026
- Google's current Docs help page tells you to check that your microphone is plugged in and is not being used by another application, when voice typing does not hear you.Google (Docs Editors Help), 2026
- Google's current Docs help page warns that punctuation might not be available in every language.Google (Docs Editors Help), 2026
- Google's current Docs help page, in its 10 October 2026 archived copy, lists Hindi among the languages and accents that voice typing works in.Google (Docs Editors Help), 2026
- Google's current Docs help page, in its 10 October 2026 archived copy, lists two Urdu entries, Urdu (India) and Urdu (Pakistan), among the languages and accents that voice typing works in.Google (Docs Editors Help), 2026
- Google's current Docs help page, in its 10 October 2026 archived copy, says voice commands are available only in English, and that both the account language and the document language must be English.Google (Docs Editors Help), 2026
- Google's Docs help page for voice typing, in its 10 October 2026 archived copy, describes voice typing as a way to type and edit a document by speaking, in Google Docs and in the speaker notes and captions of Google Slides.Google (Docs Editors Help), 2026
- Google's Docs help page, in its December 2020 archived copy, advises pausing before and after each voice command.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, gives Select paragraph, italics and Go to the end of the line as examples of spoken editing and formatting commands.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, lists Copy, Cut, Paste, Delete, Delete last word and Delete [word or phrase] among the commands to edit a document by voice.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, lists editing and formatting commands in tables such as Select text, Format your document and Edit your document.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, lists six punctuation phrases for voice typing: Period, Comma, Exclamation point, Question mark, New line and New paragraph.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, lists three steps when voice typing shows the error that it is having trouble hearing you.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, lists voice commands for tables, including Insert row, Insert new row below, Delete row and Delete table.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, says All caps makes a word or phrase all capitals, and Capitalize capitalizes only the first letter.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, says that saying Delete by itself deletes the word before the cursor.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, says the microphone shows a bubble containing the most recent command.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, says to check that the microphone is plugged in and not being used by another application.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, says you can open a list of voice commands in the document by saying a help phrase.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, says you can open a list of voice commands in the document by saying Voice typing help or Voice commands list.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, says you can say Resume with a word or phrase to move the cursor to the end of that word or phrase and start voice typing again.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, says you say Resume to move the cursor to the end of the paragraph and start voice typing again.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, says you stop voice typing by saying Stop listening.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, tells you to check that Docs heard the right command, and to say Undo if it did not.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, tells you to check your microphone settings in your computer's system preferences if the microphone does not work.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Google's Docs help page, in its December 2020 archived copy, warns that punctuation phrases might not be available in every language.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
- Handy's Ubuntu 26.04 troubleshooting guide says it was tested on Ubuntu 26.04.1 LTS with GNOME on Wayland and Handy 0.9.8.GitHub, cjpais/Handy, 2026
- If an API key or a third-party provider is set up, it takes precedence over a claude.ai sign-in, so Claude Code's docs say to remove it and restart before /voice can work.Anthropic, Anthropic, 2026
- If the Dictate button is missing in Word Mobile, Microsoft says to make sure you are signed in with an active Microsoft 365 subscription.Microsoft, 2026
- If you use the built-in microphone and voice typing does not work well, Microsoft suggests trying a headset or external microphone.Microsoft, Microsoft Support, n.d.
- In 2018, Nature described Talon and VoiceCode as popular voice programming platforms that were for Mac OS only.Julian Nowogrodzki, Nature, 2018
- In a 2022 study of programmers with motor impairments, the interviewees said they were frustrated by long hours of memorizing unnatural commands to enter code by voice.Sadia Nowrin et al., ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '22), poster, 2022
- In a 2026 preprint, students who typed their prompts were more likely than students who submitted unedited voice prompts to succeed on the first attempt, on two of three problems.Kaitlin Riegel et al., arXiv, 2026
- 58 percentIn a field study of professional developers, developers spent about 58 percent of their time on program comprehension activities.Xin Xia et al., IEEE Transactions on Software Engineering, 2018
- In a writable editor, VS Code's built-in dictation starts with the command Voice: Start Dictation in Editor.Visual Studio Code documentation (Microsoft), 2026
- In an April 2025 OpenAI community post, a user reported that ChatGPT dictation registered nothing in Chrome and, in the Android app, heard the voice but delivered no text.OpenAI Developer Community, 2025
- In Claude Code, pressing Esc or Ctrl+C cancels a dictation, discards the transcript and restores the prompt to what it held before recording.Anthropic, Anthropic, 2026
- In Claude Code's tap mode, a transcript is submitted automatically only when it is at least three words long, and shorter transcripts are inserted but not sent.Anthropic, Anthropic, 2026
- In Claude Desktop on Mac, you press Caps Lock to dictate your message to Claude with real-time transcription.Anthropic, Claude Help Center, 2026
- In Claude Desktop quick entry on Mac, you press Caps Lock once to start dictating, press it again when you finish speaking, and then click the arrow to send.Anthropic, Claude Help Center, 2026
- In Claude, dictation converts speech to text so you can type prompts by speaking, while voice mode is a two-way spoken conversation.Anthropic, Anthropic, 2026
- In Dictate in Word, saying 'delete that' removes the last spoken utterance, while saying 'delete' by itself removes only the last word or punctuation mark.Microsoft, 2026
- In Google's table of Chromebook Dictation commands, Cancel stops dictation and speech recognition.Google (Chromebook Help), 2026
- In Google's table of newer Chromebook Dictation commands, Delete all deletes all text in the text area.Google (Chromebook Help), 2026
- In Google's table of newer Chromebook Dictation commands, Repeat repeats the previous action.Google (Chromebook Help), 2026
- In Microsoft's Dictate settings, you click the gear icon to see the available settings, and Spoken Language is where you view and change the dictation language, in a drop-down.Microsoft, 2026
- In new Outlook, Microsoft says you start a new email or reply, select the body of the message, go to the Message tab and select Dictate.Microsoft, Microsoft Support, 2026
- In Outlook for Mac, Microsoft says you can start dictation with the keyboard shortcut Option + F1.Microsoft, Microsoft Support, 2026
- In Talon's command mode, speech is interpreted as commands by default, not transcribed as text.Talon community wiki (talon.wiki), n.d.
- In Talon's mixed mode, anything said that is not a command, or a chain of commands, is transcribed as plain text.Talon community wiki (talon.wiki), n.d.
- 15 seconds of silenceIn tap mode, Claude Code dictation stops recording automatically after 15 seconds of silence or two minutes in total.Anthropic, Anthropic, 2026
- 0.35 percentIn the 2022 Stack Overflow Developer Survey, 0.35 percent of respondents said they were unable to type or found it difficult to type.Stack Overflow, 2022
- In the same 2026 preprint, there was no difference in first-attempt success when students edited their transcribed voice prompts before submitting.Kaitlin Riegel et al., arXiv, 2026
- In the same original post, Andrej Karpathy wrote that he talks to Cursor's Composer with SuperWhisper, so he barely touches the keyboard.Simon Willison, Simon Willison's Weblog, 2025
- In VS Code for the Web, dictation streams microphone audio to the Microsoft AI voice service for cloud transcription, because the on-device transcription is not supported there.Visual Studio Code documentation (Microsoft), 2026
- In VS Code, built-in dictation inserts text in the chat input but does not submit the request.Visual Studio Code documentation (Microsoft), 2026
- In Wayland, the compositor sends input events to its clients.freedesktop.org, Wayland project, n.d.
- In Word Mobile, Microsoft says you start dictation by tapping the microphone icon, then wait for the Dictate button to turn on before speaking.Microsoft, 2026
- Linux Mint 22.3 has an Ubuntu Noble package base, which is the Ubuntu 24.04 LTS package set.Linux Mint, 2026
- Linux Mint's next major release, 23.0, is planned for Christmas 2026, with Cinnamon gaining fully supported Wayland sessions.Bobby Borisov, Linuxiac, 2026
- Linuxiac reports that the initial release of Canonical's Project Myna, a local dictation feature, targets Ubuntu Desktop on Wayland, with GNOME.Bobby Borisov, Linuxiac, 2026
- MDN marks the processLocally property as experimental and tells developers to check the browser compatibility table before using it in production.MDN Web Docs (Mozilla), 2026
- MDN says that on some browsers, including Chrome, speech recognition on a web page uses a server-based engine, and the audio is sent to a web service.MDN Web Docs (Mozilla), 2026
- 139 Chrome versionMDN's compatibility data lists Chrome's unprefixed SpeechRecognition from version 139, and a webkit-prefixed version from version 33.MDN Web Docs (Mozilla), GitHub, 2026
- MDN's compatibility data lists Edge's support for SpeechRecognition as a mirror of Chrome's, because Edge uses the same engine.MDN Web Docs (Mozilla), GitHub, 2026
- MDN's compatibility data lists Firefox's support for SpeechRecognition only as preview, not as a numbered stable release.MDN Web Docs (Mozilla), GitHub, 2026
- 14.1 Safari versionMDN's compatibility data lists Safari's webkit-prefixed SpeechRecognition from version 14.1.MDN Web Docs (Mozilla), GitHub, 2026
- MDN's compatibility data notes that Chrome's prefixed version needs the code to be served through a web server for recognition to work.MDN Web Docs (Mozilla), GitHub, 2026
- MDN's data guidelines use the value preview when a browser's future stable version number is unknown or uncertain.MDN Web Docs (Mozilla), GitHub, 2026
- Microsoft describes Dictate in OneNote as a quick way to create drafts, capture notes and make comments.Microsoft Support, 2026
- Microsoft describes one category of connected experiences as services that use your Microsoft 365 content to give design recommendations, editing suggestions and data insights.Microsoft, 2026
- Microsoft labels VS Code's built-in dictation as experimental, and says it might change or be removed.Visual Studio Code documentation (Microsoft), 2026
- Microsoft says a Teams transcript can be accessed through the meeting chat and the Recap tab until a meeting organizer deletes it.Microsoft, Microsoft Support, 2026
- Microsoft says an admin can run the Teams transcripts diagnostic tool in the Microsoft 365 admin center to check whether a user's account meets the requirements to transcribe a Teams meeting.Microsoft Learn, 2026
- 1 minuteMicrosoft says an audio clip recorded in Teams runs for up to one minute.Microsoft, Microsoft Support, 2026
- Microsoft says an uploaded file is transcribed in a time that depends on your internet speed, and can take up to about the length of the audio file, so a long recording may take as long as the recording itself.Microsoft Support, n.d.
- Microsoft says audio recording redirection controls whether a microphone is accessible to apps in a remote session, so a remote app cannot use your local microphone unless that redirection is allowed.Microsoft, Microsoft Learn, 2026
- Microsoft says desktop apps cannot be turned on or off one by one for the microphone, and are controlled together with the Let desktop apps access your microphone setting.Microsoft Support, 2026
- Microsoft says Dictate in Microsoft 365 is set to the language of your document by default, so you change the language only when you need another one.Microsoft, 2026
- Microsoft says Dictate in Office needs an active Microsoft 365 subscription and a version of Windows 10 or later.Microsoft Support, n.d.
- Microsoft says Dictate in PowerPoint is a quick and easy way to add content into presentation placeholders and slide notes.Microsoft Support, 2026
- Microsoft says Dictate in PowerPoint on Windows is available only to Microsoft 365 subscribers using the desktop client.Microsoft Support, 2026
- Microsoft says Dictate is not available in Office 2016 or 2019 without a Microsoft 365 subscription.Microsoft Support, 2026
- Microsoft says Dictation in Microsoft 365 uses speech-to-text with a microphone and a reliable internet connection.Microsoft, 2026
- Microsoft says Dictation in Office is set to your document language by default.Microsoft Support, 2026
- Microsoft says Dictation in Office needs a microphone and a reliable internet connection.Microsoft Support, 2026
- Microsoft says dictation in Outlook is speech-to-text that needs a microphone and a reliable internet connection.Microsoft, Microsoft Support, 2026
- Microsoft says Dictation is set to your document language in Office by default.Microsoft Support, 2026
- Microsoft says its master setting for connected experiences also turns off other services, including document co-authoring and online file storage.Microsoft, 2026
- Microsoft says languages marked as Preview in Outlook dictation might have lower accuracy or limited punctuation support.Microsoft, Microsoft Support, 2026
- Microsoft says Preview Languages may have lower accuracy or limited punctuation support.Microsoft Support, 2026
- Microsoft says some Microsoft 365 features are backed by cloud-based services, which it calls connected experiences.Microsoft, 2026
- Microsoft says some OneNote Dictate commands, marked with an asterisk, are available only in the web version of OneNote.Microsoft Support, 2026
- Microsoft says that if the Dictate button is grayed out in OneNote, you should make sure the note is not in a read-only state.Microsoft Support, 2026
- Microsoft says that if the Dictate button is grayed out, you should make sure the document is not in a read-only state.Microsoft Support, 2026
- Microsoft says that in new Outlook you start a new email or reply, select the body of the message, go to the Message tab and select Dictate.Microsoft, Microsoft Support, 2026
- Microsoft says that in Outlook 2016 you open disabled add-ins from the File tab, under Slow and Disabled COM Add-ins, and in Outlook 2013 under Slow and Disabled Add-ins.Microsoft Learn, Outlook troubleshooting, 2016
- Microsoft says that in Outlook dictation you can speak commands to add punctuation and enter special characters, in addition to dictating your content.Microsoft, Microsoft Support, 2026
- Microsoft says that in PowerPoint for the web you select Dictate on the Home tab, in the Voice section.Microsoft Support, 2026
- Microsoft says that in Word for Mac you can start dictation with the keyboard shortcut Option + F1.Microsoft, 2026
- Microsoft says that in Word you start a recording for transcription from Home, then Dictate, then Transcribe.Microsoft Support, n.d.
- Microsoft says that people who call into a Teams meeting cannot view the transcript.Microsoft, Microsoft Support, 2026
- Microsoft says that to use Dictate in PowerPoint on Windows you go to Home, then Dictate, while signed into Microsoft 365 on a device with a microphone.Microsoft Support, 2026
- Microsoft says that transcripts created without recordings can be used by Microsoft Copilot and Intelligent Recap to generate meeting insights where recording is restricted.Microsoft 365 Message Center, mirrored at mc.merill.net, 2026
- Microsoft says that when a recording is turned on but transcription is turned off, the recording does not have a transcript file stored next to it.Microsoft Learn, 2026
- Microsoft says that when a Teams meeting is over, its transcript is immediately available in the meeting event on the calendar.Microsoft, Microsoft Support, 2026
- Microsoft says that when an organizer turns off Microsoft Copilot in Teams meetings and events, recording and transcription are also turned off.Microsoft Learn, 2026
- Microsoft says that when the Who can record and transcribe option is set to No one, all automatic transcription options are forced Off.Microsoft 365 Message Center, mirrored at mc.merill.net, 2026
- Microsoft says that when you record a Teams meeting, transcription starts automatically.Microsoft, Microsoft Support, 2026
- Microsoft says that when you use Dictate in Word, your speech utterances are sent to Microsoft and used only to provide text results.Microsoft, 2026
- Microsoft says that with Teams Premium and Copilot, meeting organizers can manage recording and transcribing permissions when they schedule Teams meetings.Microsoft, Microsoft Support, 2026
- Microsoft says the audio file, whether recorded in Transcribe or uploaded, is saved to the Transcribed Files folder in OneDrive.Microsoft Support, n.d.
- Microsoft says the commands marked with an asterisk in its OneNote Dictate list work only in the web version of OneNote.Microsoft Support, 2026
- Microsoft says the Dictate feature in the Word desktop app is only available to Microsoft 365 subscribers.Microsoft, 2026
- Microsoft says the dictation feature in Outlook is only available to Microsoft 365 subscribers.Microsoft, Microsoft Support, 2026
- Microsoft says the dictation feature in Word is only available to Microsoft 365 subscribers.Microsoft, 2026
- Microsoft says the meeting policy AllowTranscription must be set to On for the organizer to see the Transcribe only option.Microsoft 365 Message Center, mirrored at mc.merill.net, 2026
- Microsoft says the Transcribe feature is currently available only on the Windows platform in OneNote for Microsoft 365.Microsoft Support, n.d.
- Microsoft says to go to Home, then Dictate, while signed into Microsoft 365 on a device with a microphone, then wait for the button to turn on before you speak.Microsoft Support, 2026
- Microsoft says to record a Teams clip, you select Actions and apps underneath the place where you type a message.Microsoft, Microsoft Support, 2026
- Microsoft says Transcribe for OneNote is not available for GCC, GCC-H and DoD customers.Microsoft Support, n.d.
- Microsoft says Transcribe in the browser needs you to be signed into Microsoft 365 and to use the new Microsoft Edge or Chrome.Microsoft Support, n.d.
- Microsoft says Transcribe in Word, unlike Dictate, does not automatically add the audio to the document. You add the transcript from the Transcribe pane.Microsoft Support, n.d.
- Microsoft says Transcribe is available to Government tenants only for Word for the web.Microsoft Support, n.d.
- Microsoft says voice typing on Windows needs an internet connection, a working microphone and the cursor in a text box.Microsoft, 2026
- Microsoft says you can fix mistakes with your keyboard without having to toggle the microphone icon off while using Dictation in PowerPoint.Microsoft Support, 2026
- Microsoft says you can turn the connected experiences off from any Microsoft 365 app, such as Word, through File > Account > Manage Settings.Microsoft, 2026
- Microsoft says you create an audio-only recording in OneNote by clicking Home, then Record Audio.Microsoft Support, n.d.
- ALT+` keyboard shortcutMicrosoft says you resume dictation in new Outlook with the keyboard shortcut ALT+` or by pressing the Mic icon in the ribbon.Microsoft, Microsoft Support, 2026
- Microsoft's accuracy advice for PowerPoint Dictate includes avoiding or eliminating background noise that may interfere with your voice.Microsoft Support, 2026
- Microsoft's connected experiences article names Dictate as one of the experiences that analyze your content.Microsoft, 2026
- Microsoft's Dictate article for OneNote applies to OneNote for Microsoft 365 for Mac, OneNote 2024, OneNote 2021, OneNote 2016 and the OneNote web app.Microsoft Support, 2026
- Microsoft's Dictate article for Word Mobile applies to Word on iPad, Word on iPhone, and Word on Android tablets and phones.Microsoft, 2026
- Microsoft's Dictate Help says that on Windows you use Alt + ` (backquote) to toggle the microphone, so a stray press of that key can turn dictation off.Microsoft Support, 2026
- Microsoft's Dictate in Microsoft 365 article lists Word, Outlook and PowerPoint under Applies To, and does not list Excel.Microsoft, Microsoft Support, 2026
- Microsoft's Dictate in Word article lists Word 2019 and Word 2016 in its Applies To list, although its troubleshooting says Dictate is not available in Office 2016 or 2019 for Windows without Microsoft 365.Microsoft, 2026
- Microsoft's Dictate in Word article lists Word 2024 and Word 2021 for Windows in its Applies To list, next to Word for Microsoft 365, although its text says the feature is only for Microsoft 365 subscribers.Microsoft, 2026
- Microsoft's Dictate your emails in Outlook article lists these products under Applies To: Outlook for Microsoft 365, Outlook for Microsoft 365 for Mac, Outlook 2024, Outlook.com, New Outlook for Windows and Outlook Web App. It lists no phone app.Microsoft, Microsoft Support, 2026
- Microsoft's English command table for Dictate in Word includes delete commands for a range, such as 'delete from [x] to [y]', and for a count, such as 'delete last [1-100] words'.Microsoft, 2026
- Microsoft's English command table for Dictate in Word includes formatting commands such as 'bold last [1-100] words' and 'bold last [1-100] sentences', and the same pattern for italicize, underline and other styles.Microsoft, 2026
- Microsoft's English command table for Dictate in Word includes list commands such as 'create bullet list', 'create number list' and 'next bullet or next item'.Microsoft, 2026
- Microsoft's English command table for Dictate in Word lists 'delete or scratch or erase', 'delete that or scratch that' and 'undo' as commands.Microsoft, 2026
- Microsoft's English command table for Dictate in Word lists 'pause dictation', 'stop dictation' and 'exit/close dictation' as dictation commands.Microsoft, 2026
- Microsoft's English punctuation table for Dictate in Word lists spoken commands for apostrophe-s, colon, semicolon, open and close quotes, hyphen and ellipsis, in addition to period, comma, question mark, exclamation mark and new line.Microsoft, 2026
- Microsoft's Excel shortcut article applies to Excel for Microsoft 365 and to Excel for Microsoft 365 for Mac.Microsoft Support, n.d.
- Microsoft's Excel shortcut table lists F2 as the key that edits the active cell and puts the insertion point at the end of its contents.Microsoft Support, n.d.
- Microsoft's message about Transcribe only says the feature was rolling out from late April to early June 2026.Microsoft 365 Message Center, mirrored at mc.merill.net, 2026
- September 2024Microsoft's message center said it would retire the Dictation feature from Outlook for iOS and Android starting early September 2024 and ending by late September 2024.Microsoft, Microsoft 365 Message Center (archived copy on mc.merill.net), 2024
- Microsoft's notice on the Outlook mobile retirement says a system microphone is still available to users while they work in the Outlook app on iOS and Android.Microsoft, Microsoft 365 Message Center (archived copy on mc.merill.net), 2024
- Microsoft's OneNote Dictate command table includes 'show help', 'pause dictation' and 'exit dictation'.Microsoft Support, 2026
- Microsoft's OneNote Dictate troubleshooting says that when there is no microphone access, you should make sure no other application or web page is using the microphone and try again.Microsoft Support, 2026
- Microsoft's OneNote Dictate troubleshooting says to make sure you have Windows 10 or higher when you cannot find the Dictate button.Microsoft Support, 2026
- Microsoft's OneNote recording article mentions a Transcribe button and says the Record button can depend on the OneNote version.Microsoft Support, n.d.
- Microsoft's OneNote recording article says that to transcribe an existing recording you should read a separate help article about transcribing in OneNote.Microsoft Support, n.d.
- Microsoft's steps for a live transcription that is not part of a recording begin by selecting More actions in your meeting controls.Microsoft, Microsoft Support, 2026
- Microsoft's steps for downloading a Teams transcript start in Chat, then open the past meeting chat, then select Recap.Microsoft, Microsoft Support, 2026
- Microsoft's Transcribe article lists the new Microsoft Edge or Chrome and an internet connection under system requirements, without saying which app each one applies to.Microsoft Support, n.d.
- Microsoft's troubleshooting for a Dictate button that cannot be found says to check that you are on the new Edge, Firefox or Chrome.Microsoft Support, 2026
- Microsoft's troubleshooting for a Dictate button that cannot be found says to make sure you are signed in with your Microsoft account.Microsoft Support, 2026
- Microsoft's troubleshooting for Dictate in Word Mobile says that if you cannot see the dictate button, you should make sure you are signed in with an active Microsoft 365 subscription.Microsoft, 2026
- Microsoft's troubleshooting for Dictate in Word says that if Dictation cannot hear you, or nothing appears as you dictate, you should make sure your microphone is not muted and adjust its input level.Microsoft, 2026
- Microsoft's troubleshooting for Outlook dictation says that if nothing appears as you dictate, you should make sure your microphone is not muted.Microsoft, Microsoft Support, 2026
- Microsoft's troubleshooting for Word says that if you see a message about microphone access, you should refresh the page, click Dictate and give the browser permission to use the microphone.Microsoft, 2026
- Microsoft's troubleshooting says to make sure no other application or web page is using the microphone when Dictate reports that it has no access.Microsoft Support, 2026
- Microsoft's voice typing troubleshooting says to make sure the microphone you want to use is selected in Settings.Microsoft, Microsoft Support, n.d.
- Mozilla's wiki says that speech input from Chrome can be saved and attached to a user's profile, which its own requests sent to Google are not.Mozilla, wiki.mozilla.org, 2026
- Nature's 2018 feature explains that coding by voice requires a speech-recognition engine and a separate platform for voice coding.Julian Nowogrodzki, Nature, 2018
- Notion announced on April 6, 2026 that voice input for Notion AI was now available on desktop.Notion, Notion Releases, 2026
- 10 hours per user per dayNotion says AI Meeting Notes has a daily usage limit of 10 hours per user.Notion, Notion Help Center, 2026
- Notion says AI Meeting Notes requires a Business or Enterprise plan, with an exception for eligible mobile subscriptions.Notion, Notion Help Center, 2026
- 1 minuteNotion says at least one minute of recorded audio is required to generate an AI Meeting Notes summary.Notion, Notion Help Center, 2026
- Notion says the browser version of AI Meeting Notes can only pick up audio from the microphone.Notion, Notion Help Center, 2026
- 4.7.0 app versionNotion says the desktop app must be on version 4.7.0 or higher to use AI Meeting Notes.Notion, Notion Help Center, 2026
- Notion's April 6, 2026 release describes its desktop voice input as a way to dictate longer prompts or talk instead of typing, to get things done with the Notion Agent.Notion, Notion Releases, 2026
- Notion's pricing page describes its meeting notes feature as one that automatically transcribes your meetings, along with a helpful summary.Notion Labs, Inc., 2026
- Nuance says offline mode is not supported for working with web-based documents in the Dragon Web Extension, so dictating into a web app needs a connection.Nuance Communications, Dragon for PC help (version 15), 2017
- Nuance's Dragon 15 help says the Dragon Web Extension does not apply to Microsoft Edge, so you cannot click web page objects by voice in Edge.Nuance Communications, Dragon for PC help (version 15), 2017
- Nuance's Dragon 15.3 release notes list as fixed that Dragon no longer causes Office applications to fail, and that dictation no longer stops working in Office applications.Nuance Communications, Dragon Professional Individual help, 2017
- Office 2013 and later versions disable an add-in that performs slowly, and Microsoft says you can re-enable it with the Always enable this add-in option.Microsoft Learn, Outlook troubleshooting, 2016
- Office Watch reported in July 2024 that Microsoft would drop Dictation from Outlook on iPhone, iPad and Android for no reason Microsoft would share.Office-Watch.com, Office Watch, 2024
- On a Mac, Microsoft says you go to Home, then Dictate, while signed into Microsoft 365 on a device with a microphone.Microsoft Support, 2026
- On a supported desktop platform, VS Code downloads a speech recognition model on first use, and after that speech recognition works locally and offline.Visual Studio Code documentation (Microsoft), 2026
- On iOS, ChatGPT automatically retries dictation once if it fails because of a connection issue or timeout.OpenAI, OpenAI Help Center, 2026
- On its OneNote page, Microsoft says Dictate is not available in Office 2016 or 2019 for Windows without Microsoft 365.Microsoft Support, 2026
- On January 12, 2026, OpenAI said it was improving ChatGPT dictation for all logged-in users and significantly reducing empty transcriptions.OpenAI, OpenAI Help Center, 2026
- On the desktop, VS Code's built-in dictation uses an on-device speech recognition model by default, so dictation does not need to send audio to an online service.Visual Studio Code documentation (Microsoft), 2026
- On Windows, a program may send simulated input only to applications at an equal or lower integrity level, a restriction called User Interface Privilege Isolation (UIPI).Microsoft, Microsoft Learn, Windows application development documentation, n.d.
- On Windows, Wispr Flow's docs say it uses Shift+Insert instead of Ctrl+V in supported coding editors, including their integrated terminals when needed.Wispr Flow help center, 2026
- OpenAI Academy says that in the ChatGPT mobile or desktop app you start dictation by tapping the microphone icon in the text box, and your words then appear as text.OpenAI, OpenAI Academy, 2026
- OpenAI says audio from ChatGPT dictation is kept for as long as the chat stays in your chat history, and that deleting the chat removes the audio clip within 30 days, with exceptions for security or legal reasons.OpenAI, OpenAI Help Center, 2026
- OpenAI says that when you press the microphone icon in ChatGPT, the recorded audio is sent to its models to be transcribed, and the returned text can be edited before you send it.OpenAI, OpenAI Help Center, 2026
- OpenAI says voice dictation is for turning speech into prompt text before you send it, and is separate from ChatGPT Voice, which is a live spoken conversation.OpenAI, OpenAI Developers, 2026
- 10 percentOpenAI says word error rate in its ChatGPT dictation evaluations was at least 10% lower for the top languages tested than with the previous production model.OpenAI, OpenAI Help Center, 2026
- OpenAI's ChatGPT Voice help article says to use ChatGPT Dictation when you want to record a prompt, review and edit its transcription, and then send it as text.OpenAI, OpenAI Help Center, 2026
- OpenAI's Codex configuration has a features.in_app_dictation setting that turns off dictation in the Codex desktop app.OpenAI developer documentation, 2026
- OpenAI's dictation FAQ says that consumer users who have chosen to share audio to improve its models may have their dictation audio used for training.OpenAI, OpenAI Help Center, 2026
- OpenAI's release notes say it began rolling out a new speech-to-text model for dictation in ChatGPT across all plans, with no change to how users dictate.OpenAI, OpenAI Help Center, 2026
- OpenAI's speech-to-text guide names whisper-1 as the model to use for translating a completed audio recording into English.OpenAI, OpenAI, 2026
- OpenAI's speech-to-text guide shows the transcription quickstart sending audio to the transcriptions endpoint with the model gpt-transcribe.OpenAI, OpenAI, 2026
- Serenade's documentation says the app is freely available for macOS, Windows and Linux.Serenade, n.d.
- Serenade's download page says the app works across Mac, Windows and Linux.Serenade, n.d.
- Serenade's home page describes its speech-to-code engine as designed for developers and fully open-source.Serenade, n.d.
- Serenade's home page says that with Serenade you can write code using natural speech.Serenade, n.d.
- Serenade's home page says the app can process speech in the cloud or completely locally, so that voice commands and source code stay on the device.Serenade, n.d.
- Simon Willison defines vibe coding as building software with an LLM without reviewing the code it writes, and he says that using LLMs for code responsibly is not vibe coding.Simon Willison, Simon Willison's Weblog, 2025
- Slack adds a transcript to a recorded audio clip when you click the Generate transcript icon.Slack, Slack Help Center, 2026
- Slack says AI notes cannot be used in a huddle that includes external people or guests.Slack, Slack Help Center, 2026
- Slack says AI notes for huddles are available on paid plans, and that all members can use them but not guests.Slack, Slack Help Center, 2026
- 5 minutesSlack says audio and video clips can be up to five minutes long.Slack, Slack Help Center, 2026
- Slack says huddle transcripts are embedded in the notes canvas and do not show up in search results.Slack, Slack Help Center, 2026
- Slack's help center says the Slackbot prompt box has a microphone icon that dictates a prompt instead of typing it.Slack, Slack Help Center, 2026
- Some keyboard keys are not delivered to terminal applications, so Claude Code's dictation key cannot be set to them.Anthropic, Anthropic, 2026
- Speech Note's README gives Flatpak from Flathub as the Linux install route, using the package name net.mkiol.SpeechNote.GitHub, mkiol/dsnote, 2026
- Speech Note's README lists Vosk and whisper.cpp with Parakeet among the speech to text engines it uses.GitHub, mkiol/dsnote, 2026
- 1000 commandsTavis Rudd's spoken Python language required users to learn a complex set of over 1000 commands.Sadia Nowrin et al., ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '22), poster, 2022
- The 2022 paper says Serenade requires that spoken commands precisely match the commands the system understands.Sadia Nowrin et al., ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '22), poster, 2022
- The 2022 paper says Serenade was created out of need when a co-founder was diagnosed with RSI.Sadia Nowrin et al., ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '22), poster, 2022
- The 2026 arXiv preprint Lost in Transcription names custom identifiers, such as variable and function names, as a central difficulty for speech input about code.Jayant Havare et al., arXiv, 2026
- The activeTab permission gives an extension temporary access to the current tab only when the user invokes the extension, such as by clicking its action.Chrome for Developers, 2026
- 74 million parametersThe base Whisper model has 74 million parameters.OpenAI, GitHub (OpenAI repository), 2023
- The Caps Lock dictation shortcut in Claude Desktop quick entry is disabled by default, because it takes over the Caps Lock key.Anthropic, Claude Help Center, 2026
- 10,000+ sitesThe Chrome Web Store listing for the Voice In extension says it provides voice typing in ChatGPT, Gmail, Google Docs and more than 10,000 sites, in 50 or more languages.Chrome Web Store (developer listing, Voice In), 2026
- The Claude Code VS Code extension's voice dictation is not available in VS Code Remote sessions, including SSH, Dev Containers and Codespaces.Anthropic, Anthropic, 2026
- The community command set for Talon lists Talon's built-in Conformer (wav2letter) speech recognition engine as the recommended engine, with Dragon as the alternative.talonhub, GitHub, n.d.
- The current Ubuntu 26.04 release can also try out Myna, Canonical's voice typing app.UbuntuHandbook, 2026
- The Cursorless installation guide starts with installing Talon and a speech engine, then the community Talon commands and the VS Code extension.Cursorless documentation, n.d.
- The developer of Talon left a full-time software engineering job and started building Talon after developing severe hand pain.Sadia Nowrin et al., ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '22), poster, 2022
- The Dictation for Gmail listing says that to dictate, you open a new message or reply to one, then click the microphone icon to the right of the Send button.Chrome Web Store (developer listing, Dictation for Gmail), 2026
- 3.0 out of 5 average rating from 274 ratingsThe Dictation for Gmail store listing shows an average rating of 3.0 out of 5 from 274 ratings.Chrome Web Store (developer listing, Dictation for Gmail), 2026
- January 1, 2025 last update dateThe Dictation for Gmail store listing shows version 1.0.14, last updated on January 1, 2025.Chrome Web Store (developer listing, Dictation for Gmail), 2026
- The fix in Handy's Ubuntu 26.04 GNOME Wayland guide is to type with ydotool and to handle shortcuts with handy_keys.GitHub, cjpais/Handy, 2026
- 1550 million parametersThe large Whisper model has 1,550 million parameters.OpenAI, GitHub (OpenAI repository), 2023
- 39.7 times as many parametersThe large Whisper model has about 40 times as many parameters as the tiny model.Calculated from other facts
- 10 GB of video memoryThe large Whisper model needs about 10 GB of video memory according to OpenAI's model table.OpenAI, GitHub (OpenAI repository), 2023
- 769 million parametersThe medium Whisper model has 769 million parameters.OpenAI, GitHub (OpenAI repository), 2023
- July 23, 2022 date of latest commit on the main branchThe most recent commit on the main branch of Serenade's public GitHub repository is dated July 23, 2022.GitHub (Serenade public repository), 2022
- The Myna client and its AI models are installed as snap packages.UbuntuHandbook, 2026
- The nerd-dictation project describes itself as offline speech to text for desktop Linux, built on the VOSK-API.Campbell Barton, GitHub, 2026
- The nerd-dictation README lists xdotool, a command to simulate input in X11, as its default input simulation utility.Campbell Barton, GitHub, 2026
- The not-allowed error means the browser blocked speech input for security, privacy or user preference reasons.MDN Web Docs (Mozilla), 2026
- The privacy section of the Dictation for Gmail store listing says the extension handles personally identifiable information, and that the developer declares the data is not being sold.Chrome Web Store (developer listing, Dictation for Gmail), 2026
- The privacy section of the Speech Recognition Anywhere listing lists personally identifiable information among the data the developer handles.Chrome Web Store, 2026
- The privacy section of the Speech to Text listing lists personally identifiable information and authentication information among the data the developer handles.Chrome Web Store, 2026
- The privacy section of the Speechnotes listing says the developer has disclosed that it will not collect or use user data.Chrome Web Store, 2026
- The programmers with motor impairments interviewed in the 2022 study all used a combination of mouse, keyboard and voice to program.Sadia Nowrin et al., ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '22), poster, 2022
- 244 million parametersThe small Whisper model has 244 million parameters.OpenAI, GitHub (OpenAI repository), 2023
- The specification defines a downloadable status for on-device recognition that the browser supports but that is not yet installed.W3C Web Speech API Community Group, 2026
- 30000 usersThe Speech Recognition Anywhere Chrome Web Store listing shows 30,000 users, checked on 2026-10-17.Chrome Web Store, 2026
- The Speech Recognition Anywhere listing describes the extension as expanding the Web Speech API in Chrome and Edge.Chrome Web Store, 2026
- 1000 usersThe Speech to Text Chrome Web Store listing shows 1,000 users, checked on 2026-10-17.Chrome Web Store, 2026
- The Speech to Text listing says all data and voice transcription are handled locally or through encrypted channels.Chrome Web Store, 2026
- 20000 usersThe Speechnotes Chrome Web Store listing shows 20,000 users, checked on 2026-10-17.Chrome Web Store, 2026
- The Speechnotes listing says it uses Google's speech recognition engine for Chrome and that it does not collect or share user data.Chrome Web Store, 2026
- 39 million parametersThe tiny Whisper model has 39 million parameters.OpenAI, GitHub (OpenAI repository), 2023
- The W3C TAG review of the on-device Web Speech feature names the potential for fingerprinting as its primary risk.W3C Technical Architecture Group (review request), 2025
- The Web Speech API specification is a Community Group draft report dated 15 October 2026, not a finished W3C standard.W3C Web Speech API Community Group, 2026
- The Web Speech API specification requires browsers to start a speech input session only with explicit, informed user consent.W3C Web Speech API Community Group, 2026
- 10 GB of VRAMThe Whisper README gives about 10 GB of required VRAM for the large model, which the table rates as the 1x speed baseline.OpenAI, GitHub (OpenAI repository), 2023
- 2 GB of VRAMThe Whisper README gives about 2 GB of required VRAM for the small model.OpenAI, GitHub (OpenAI repository), 2023
- 5 GB of VRAMThe Whisper README gives about 5 GB of required VRAM for the medium model.OpenAI, GitHub (OpenAI repository), 2023
- 10 times the relative speed of largeThe Whisper README gives the tiny model a relative speed of about 10 times the large model, which it rates 1x.OpenAI, GitHub (OpenAI repository), 2023
- The whisper.cpp project lists support for CPU-only inference among its features, alongside GPU support for NVIDIA, AMD and Vulkan.Georgi Gerganov, GitHub, 2026
- The xdotool README says that typing does not work on Wayland, and that it is unclear whether it could ever work there.Jordan Sissel, GitHub, jordansissel/xdotool, 2026
- To dictate a chat prompt in VS Code, select the microphone button in the chat input.Visual Studio Code documentation (Microsoft), 2026
- Transformers.js runs a model on the CPU through WebAssembly by default in the browser, and on the GPU through WebGPU only when the page sets the device option.Hugging Face, GitHub, 2026
- Visual Studio Code's voice documentation says its built-in dictation converts speech to text in chat, the Agents window, editors and terminals.Visual Studio Code documentation (Microsoft), 2026
- Vocalinux describes itself as a free, AGPL-3.0-licensed desktop app for X11 and Wayland.GitHub, VocaHQ/vocalinux, 2026
- Vocalinux says it types recognized text on X11 and Wayland through xdotool, IBus, wtype, ydotool or a clipboard fallback.GitHub, VocaHQ/vocalinux, 2026
- Vocalinux turns speech into typed text in whatever app has focus, which makes it a dictation tool rather than a voice command tool.GitHub, VocaHQ/vocalinux, 2026
- Voice dictation in Claude Desktop quick entry needs macOS 14 or later.Anthropic, Claude Help Center, 2026
- VS Code's built-in dictation converts speech to text in chat, the Agents window, editors and terminals.Visual Studio Code documentation (Microsoft), 2026
- VS Code's built-in dictation is not available on Intel-based Mac computers, so those users need the VS Code Speech extension instead.Visual Studio Code documentation (Microsoft), 2026
- Wayland is a new display server and compositing protocol, and Weston is the implementation of that protocol.freedesktop.org, Wayland project, n.d.
- WebKit says Safari supports speech recognition through the Web Speech API, powered by the same speech engine as Siri.WebKit (Apple), 2021
- WebKit says Safari users need Siri turned on for the speech recognition part of the Web Speech API to be available.WebKit (Apple), 2021
- 50 languages and dialectsWebKit says Safari's speech recognition, powered by the Siri engine, gives high-quality transcription for over 50 languages and dialects.WebKit (Apple), 2021
- When a page sets processLocally to true but the correct on-device language pack is not installed, start() fails with a language-not-supported error.MDN Web Docs (Mozilla), 2026
- When a page sets processLocally to true, the specification requires that recognition run on the user's device.W3C Web Speech API Community Group, 2026
- When a site wants your microphone, Chrome on a computer asks you to choose Allow while visiting the site, Allow this time, or Never allow.Google (Chrome Help), 2026
- When ChromeOS Dictation recognizes a spoken command, it runs the command as well as it can. Otherwise it types the recognized words as they are.Chromium Project (Google), undated
- When Microsoft's Dictate troubleshooting reports that it cannot detect your microphone, it says to make sure the microphone is plugged in and to test it.Microsoft, 2026
- When Outlook dictation misses words or gets many words wrong, Microsoft says to make sure you are on a fast and reliable internet connection.Microsoft, Microsoft Support, 2026
- When Outlook dictation shows 'Mic issue', 'Can't detect mic' or 'Can't access mic', Microsoft says to make sure the microphone is plugged in and to test it.Microsoft, Microsoft Support, 2026
- When Outlook dictation shows the message 'Another app is using your mic', Microsoft says to make sure no other application or web page is using the microphone, then try again.Microsoft, Microsoft Support, 2026
- When processLocally is false, which is the default, the specification lets the browser choose between local and remote processing.W3C Web Speech API Community Group, 2026
- When speech recognition fails because a network connection it needs is missing, the page receives the network error code.MDN Web Docs (Mozilla), 2026
- When the on-device SODA component is not available, ChromeOS Dictation falls back to network speech recognition.Chromium Project (Google), undated
- whisrs describes its speech to text as working on Wayland, X11, Hyprland, Sway, Niri, GNOME and KDE.docs.rs, 2026
- whisrs offers cloud transcription through Groq, Deepgram or OpenAI, and also fully offline transcription through whisper.cpp.docs.rs, 2026
- Windows' SendInput function inserts the events it is given one after another into the keyboard or mouse input stream, so typed text is not interleaved with keys the user presses at the same moment.Microsoft, Microsoft Learn, Windows application development documentation, n.d.
- Wispr Flow's help docs say that on a Mac, in Claude Code and Codex CLIs, it pastes long dictations in parts.Wispr Flow help center, 2026
- ydotool types by emulating an input device through the Linux kernel's uinput framework, and its daemon needs access to /dev/uinput.ReimuNotMoe, GitHub, ReimuNotMoe/ydotool, 2026