[BUG] Voice mode on Linux only transcribes first ~2 words before cutting off

Resolved 💬 5 comments Opened Mar 13, 2026 by janbaer Closed Apr 19, 2026

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet
  • [x] This is a single bug report (please file separate reports for different bugs)
  • [x] I am using the latest version of Claude Code

What's Wrong?

Description

Voice mode (/voice) only captures approximately the first two words of speech, then stops transcribing as if the recording ended. This happens consistently regardless of speaking pace or volume.

## Environment

  • Claude Code version: 2.1.75
  • OS: NixOS 25.11 (build 25.11.20260304.fabb8c9), kernel 6.12.74
  • Audio: PipeWire 1.4.9 (default clock rate: 48000 Hz)
  • Microphone: Working correctly in all other applications

## Steps to reproduce

  1. Start Claude Code
  2. Enter /voice to enable voice mode
  3. Hold Space and speak a full sentence
  4. Release Space
  5. Only the first ~2 words are transcribed

## Expected behavior

The full spoken sentence should be transcribed.

## Actual behavior

Only the first ~2 words are captured. The rest of the speech is lost, suggesting the Voice Activity Detection (VAD) triggers an early cutoff.

## Additional context

  • PipeWire configuration is standard (48000 Hz sample rate)
  • The microphone works correctly in other applications
  • This may be related to how VAD interacts with PipeWire on Linux

What Should Happen?

The full spoken sentence should be transcribed.

Error Messages/Logs

Only the first ~2 words are captured. The rest of the speech is lost, suggesting the Voice Activity Detection (VAD) triggers an early cutoff.

Steps to Reproduce

  1. Start Claude Code
  2. Enter /voice to enable voice mode
  3. Hold Space and speak a full sentence
  4. Release Space
  5. Only the first ~2 words are transcribed

Claude Model

Sonnet (default)

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.75

Platform

Anthropic API

Operating System

Other Linux

Terminal/Shell

Other

Additional Information

  • PipeWire configuration is standard (48000 Hz sample rate)
  • The microphone works correctly in other applications
  • This may be related to how VAD interacts with PipeWire on Linux

View original on GitHub ↗

This issue has 5 comments on GitHub. Read the full discussion on GitHub ↗