[feat]: Voice input support for claude code (microphone integration)

[maxonary]

Feature Summary

Add support for voice input so claude commands in Emdash can be dictated via the system microphone, similar to the voice functionality available in Cloud Code.

The goal is to allow users to trigger slash commands or enter prompts using speech instead of typing. Emdash should request microphone access from the operating system and transcribe spoken input directly into the command input field.

Problem or Use Case

Many users interact with coding agents using natural language. Typing longer prompts or commands can slow down workflows, especially when users are brainstorming or describing complex tasks.

Cloud Code already supports voice input via /voice and holding space. Having the same capability in Emdash would allow users to quickly dictate instructions such as:

Proposed Solution

Implement microphone input support directly in the slash command prompt:

1. Add a microphone button next to the command input field.

2. When clicked, Emdash requests microphone permission from the operating system.

3. Spoken input is captured and transcribed using either:

   • the system speech recognition APIs, or
   • a configurable speech-to-text provider.

4. The transcription appears in the slash command input field so the user can edit before submitting.

Optional enhancements:

• push-to-talk keyboard shortcut
• live transcription preview
• configurable STT backend

Alternatives Considered

No response

Additional Context

No response

[shreyaspapi]

+1

It would also be interesting to see support for native/local models (like Whisper for STT), possibly use https://github.com/RunanywhereAI/runanywhere-sdks

[arnestrickmann]

Will be shipped!