Repository navigation
Audio input instead of typing #1982
Description
Activity
I use dictation inbuilt for mac and i added keyboard shortcut once i press Control keys twice it shows the audio and I do STT.
https://support.apple.com/en-uz/guide/mac-help/mh40584/macI have not tried it on windows but you can check this https://support.microsoft.com/en-us/windows/use-voice-typing-to-talk-instead-of-type-on-your-pc-fec94565-c4bd-329d-e59a-af033fa5689f
Reacted by Daco HarkesYou can get very high quality voice to text by using a 3rd party with OpenAI whisper large integration.
Hey @nshiab! 👋
Great suggestion! I've got some exciting news - Voice Mode already provides exactly what you're looking for with Gemini CLI!
Voice Mode + Gemini CLI = 🎤✨
Check out our demo video showing Voice Mode working seamlessly with Gemini CLI!
What you get:
✅ Full voice conversations - Talk to Gemini, hear responses back
✅ Natural speaking - No push-to-talk, just speak naturally
✅ Smart silence detection - Automatically knows when you're done
✅ High-quality STT - Uses OpenAI Whisper (as @xied-ctrl suggested!) or local alternativesQuick setup:
# Set your OpenAI API key (for STT/TTS) export OPENAI_API_KEY=your-key # Configure Voice Mode in your .gemini/settings.json # See our full Gemini CLI setup guide: # https://github.com/mbailey/voicemode/blob/main/docs/integrations/gemini-cli/README.md # Start Gemini CLI gemini # Then type 'converse' to start voice mode
Then just say: "Let's have a voice conversation" and you're speaking with Gemini! 🎙️
Why it's better than OS dictation:
As @olufuwatayo mentioned, OS dictation works, but Voice Mode offers:
- Two-way voice - Hear Gemini's responses too
- Context-aware - Gemini can ask clarifying questions verbally
- Faster workflow - No switching between dictation and reading
Bonus features:
- Local STT option - Use Whisper.cpp for privacy
- Multiple voices - Choose how Gemini sounds
- Emotional speech - Gemini can express excitement, concern, etc.
Speaking really is faster than typing - and with Voice Mode, the entire conversation flows naturally!
Would love to hear your feedback after trying it out!
Reacted by Nael Shiab, Hudson Newey, xied-ctrl, Jacob Richman and Riccardo CarlessoVery interesting! Thank you all for these suggestions!
- added a parent issue
on Jul 30, 2025 - removed a parent issue
on Jul 31, 2025 - added a parent issue
on Jul 31, 2025 Related work:
https://github.com/mbailey/voicemodeReacted by Devansh Varshney (देवांश वार्ष्णेय)4 remaining items
I have been working on this
using MCP + custom slash commands
Commands
https://github.com/automateyournetwork/GeminiCLI_Slash_TalkMatching MCP
https://github.com/automateyournetwork/Talk_MCP_for_Gemini_CLIVideo in action:
https://youtu.be/beVxnNnEmroReacted by alingse- marked Bidirectional voice / audio mode #6929 as a duplicate of this issue
on Aug 26, 2025 My PR #18499 introduces an experimental version of voice input. It handles recording and transcription within the CLI application.
- marked Continuous voice guided development #21372 as a duplicate of this issue
on Mar 9, 2026 Assigned area/agent based on content keywords. Marked as kind/customer-issue with priority/p3.
What would you like to be added?
Amazing work.
But I would love to be able to talk with the agent instead of typing all the time.
I use Copilot in VS Code most of the time, and it's a feature that I love about it.
Thanks. :)
Why is this needed?
Speaking is often faster than typing.
Additional context
No response