/voice does not recognize Japanese speech — always transcribes as English

Resolved 💬 3 comments Opened Mar 13, 2026 by barewalker Closed Mar 17, 2026

Description

The /voice command consistently misrecognizes Japanese speech as English. Every attempt to speak Japanese results in garbled English transcription.

Steps to reproduce

  1. Enable voice mode with /voice
  2. Speak in Japanese (hold Space to record)
  3. Observe that the transcription is always in English, not Japanese

Expected behavior

Japanese speech should be recognized and transcribed as Japanese text.

Actual behavior

All Japanese speech is transcribed as nonsensical English words/phrases.

Environment

  • OS: Arch Linux (also WSL2)
  • Claude Code: latest
  • Language: Japanese (ja)

Feature request

Please add a manual language setting for voice recognition, e.g.:

{
  "voice:language": "ja"
}

The current auto-detection does not work for Japanese. A manual override would solve this issue.

Supported languages (per docs)

The documentation states that 20 languages including Japanese are supported, but auto-detection fails to identify Japanese speech in practice.

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗