feat(desktop): add voice input

[bernardokcosta]

Issue for this PR

Closes #XXXX

Type of change

• Bug fix
• New feature
• Refactor / code improvement
• Documentation

What does this PR do?

Adds optional voice input to the Desktop prompt composer.

Voice can be transcribed locally with a pinned whisper.cpp runtime and downloaded Whisper models, keeping recordings on the device. Users can also select a configured AI model that supports audio input.

The change includes microphone permissions, cancellation handling, model integrity checks, voice settings, packaging support, SDK generation, and translations for all supported locales.

How did you verify your code works?

• Ran typechecks for the app, desktop, server, and SDK packages
• Ran the focused app, desktop, HTTP API, SDK, and Prompt Input V2 tests
• Verified i18n parity for all locales
• Built the production Desktop application
• Compiled the pinned whisper.cpp runtime on Ubuntu x64
• Launched the Electron development application on Ubuntu

Screenshots / recordings

2026-08-11.19-09-04.mp4

Checklist

• I have tested my changes locally
• I have not included unrelated changes in this PR

[github-actions]

Thanks for updating your PR! It now meets our contributing guidelines. 👍

[github-actions]

The following comment was made by an LLM, it may be inaccurate:

Potential Duplicate PRs Found

I found several related PRs that may be addressing voice input features:

1. feat(opencode): add voice input (feat(opencode): add voice input #29663) - feat(opencode): add voice input #29663

   • Directly related to voice input functionality, though scoped to opencode rather than desktop

2. feat: add first-party voice transcription with local Whisper (feat: add first-party voice transcription with local Whisper #11345) - feat: add first-party voice transcription with local Whisper #11345

   • Implements voice transcription with Whisper, likely a related infrastructure feature

3. feat: Add voice input using browser speech recognition (web only) (feat: Add voice input using browser speech recognition (web only) #18225) - feat: Add voice input using browser speech recognition (web only) #18225

   • Voice input for web, may overlap with desktop voice input implementation

These PRs appear to be addressing similar voice input functionality across different packages/platforms. You may want to check their current status and whether PR #41864 duplicates or should be consolidated with any of these existing work items.