Feature request: Add voiceLanguage setting for /voice mode #31724

Feature Request

Problem

The /voice command currently has no language configuration option. The speech-to-text engine appears to default to English, making it unreliable for non-English languages like Ukrainian.

When speaking Ukrainian in voice mode, the transcription either fails or produces incorrect English text, even though Claude can perfectly understand typed Ukrainian input.

Proposed Solution

Add a voiceLanguage setting (e.g., in settings.json) that allows users to specify one or multiple STT languages:

{
  "voiceEnabled": true,
  "voiceLanguage": ["en", "uk"]
}

• Accept a single language code ("en") or an array of languages (["en", "uk"]) for multilingual users
• Use ISO 639-1 language codes (e.g., uk for Ukrainian, ja for Japanese, es for Spanish)
• "auto" for auto-detection (current behavior / default)

Use Case

Many developers are multilingual and switch between languages naturally during a conversation. For example, speaking both Ukrainian and English in the same session. The STT engine should be able to handle multiple specified languages rather than defaulting to English only.

Alternatives Considered

• Auto-detection: The current auto-detection doesn't reliably identify non-English languages
• Typing instead of speaking: Works, but defeats the purpose of voice mode

Additional Context

• Currently the only voice-related setting is voiceEnabled (confirmed by inspecting the binary — no voiceLanguage key exists)
• The underlying STT engines (e.g., Whisper) already support multiple languages — this just needs to be exposed as a config option