Claude Cowork voice: hear its replies and answer out loud
SpeakApp reads each Claude Cowork reply aloud, one after another, and you say “ok” to answer the session that spoke. It works with Cowork running locally in Claude Desktop, for listening and for answering by voice. Dictation types your words at the cursor and is a separate function from the reading loop. Hearing a reply does not replace checking the result on screen.
How do you answer Claude Cowork by voice?
Replies are read aloud one after another, as soon as they are ready. You say “ok”, speak your answer, then say “enter”, and your message goes straight into the session that spoke, without hunting for its window, touching the keyboard or looking at the screen.
Each conversation gets a card in the toast, the panel SpeakApp shows above your windows, with read, skip, later, reply and open. The widget's circle counts what is still waiting, and the Control Center shows which session is working, done, waiting or stuck.
Suppose a Claude Cowork task and a Claude Desktop Code session are both running. The Cowork reply arrives first and is read aloud. You say “ok”, ask for a shorter summary, then say “enter”. The message goes to the Cowork task, not to whichever window you touched last. The Code session's reply waits in the queue until you want it.
What works with Claude Cowork, and what does not?
SpeakApp listens to Claude Cowork and answers it by voice when Cowork runs locally in Claude Desktop. Listening and answering are two separate lists.
You answer by voice, straight into the session, in Claude Desktop (the Code tab and local Cowork), in the Codex app for Windows and with ChatGPT in Chrome, through the SpeakApp extension. Listening also works in Cursor and in VS Code, and with Claude.ai in Chrome, through the extension. There, you dictate your reply into the application's field. The app-by-app list is on the Features page.
One limit: a conversation that runs remotely in the cloud leaves no trace on your disk, so SpeakApp can neither read it nor reply to it.
Reading loop, dictation or keyboard?
| Option | What it does | What you still do yourself |
|---|---|---|
| Reading loop | Reads each finished reply aloud and sends your spoken answer into the session that spoke | Check results that need your eyes |
| Dictation | Types your words at the cursor, in any Windows text field | Choose the conversation and handle its replies |
| Keyboard and screen | You read the reply on screen, find the right window and type the next instruction | Watch for replies and switch between windows |
Dictation writes wherever your cursor is, so it also works in the message field of Cursor, VS Code or Claude.ai. Hearing a reply does not replace checking the result: go back to the screen when something needs a close look.
Does Claude Cowork have its own voice mode?
According to Anthropic's help page, Claude's voice mode is a spoken conversation with Claude, in beta on Claude Mobile (iOS and Android), in Claude Desktop and on the web: you speak, Claude speaks back, and you stay inside that one conversation. In Claude Cowork and Claude Code, the same page says dictation is available but voice mode isn't.
SpeakApp does something else. It reads the replies of the sessions you already have, one after another, as they finish. You can work elsewhere, and you answer by voice in the session that spoke.
Before you rely on it
Look at the conversation the toast shows before you say “ok”: the reply goes to that conversation, not to the window in front of you. The first time, check that your message landed in the right session.
Related guides
- How the reading loop works
- Claude voice alternative for parallel AI sessions
- Claude Code voice mode: what /voice does, and how to hear the replies
Frequently asked questions
Can I use Claude Cowork by voice?
With SpeakApp, yes, when Cowork runs locally in Claude Desktop. Each finished reply is read aloud. Say “ok”, speak your answer, then say “enter”: your message goes into the Cowork session that spoke.
What do I say to answer a Claude Cowork reply?
Say “ok” to start a reply to the conversation shown in the toast, speak your answer, then say “enter” to send it.
Can I follow several sessions at once?
Yes. Replies from several sessions wait in a queue and are read one after another. Each conversation has its card in the toast, and the widget's circle counts what is still waiting.
Does hearing a reply mean I can answer in every application?
No. Answering by voice straight into the session works in Claude Desktop (the Code tab and local Cowork), in the Codex app for Windows and with ChatGPT in Chrome. In Cursor, VS Code and Claude.ai you listen, and you dictate your reply into the field.
Is dictation the same as answering a session?
No. Dictation types your words at the cursor, in any Windows text field. Answering a session sends your spoken message into the conversation that spoke.
Can I see which Claude Cowork session is working or done?
Yes. The Control Center shows which session is working, done, waiting or stuck. It covers Claude Cowork running locally in Claude Desktop.
Does Claude Cowork have a voice mode of its own?
According to Anthropic's help page, dictation is available in Claude Cowork but voice mode isn't. Voice mode is a spoken conversation with Claude: you speak and Claude speaks back. SpeakApp reads the replies of Claude Cowork running locally in Claude Desktop, one after another, and lets you answer by voice in the session that spoke.
Try SpeakApp
SpeakApp is available on Windows, coming soon to Mac. Download SpeakApp or join the Mac waiting list. To see the whole product first: What is SpeakApp?