feat(voice-server): Add Google Cloud TTS as alternative provider

[fayerman-source]

Summary

Adds Google Cloud Text-to-Speech as a second TTS backend alongside ElevenLabs:

• Provider selection via settings.json → daidentity.ttsProvider ("elevenlabs" or "google-cloud")
• Backwards compatible — defaults to ElevenLabs when ttsProvider is not set
• No new dependencies — uses Google's REST API directly via fetch (no SDK)
• Configurable voice via daidentity.googleCloudVoice (language, voice name, type, rate, pitch)

Why Google Cloud TTS

• Free tier: 4M characters/month (Standard) or 1M (WaveNet) vs ElevenLabs' 10K
• No attribution requirement on the free tier
• Good fallback when ElevenLabs quota runs out

Configuration

Add to ~/.env:

GOOGLE_CLOUD_API_KEY=your_key_here

Add to ~/.claude/settings.json:

{
  "daidentity": {
    "ttsProvider": "google-cloud",
    "googleCloudVoice": {
      "languageCode": "en-US",
      "voiceName": "en-US-Neural2-D",
      "voiceType": "NEURAL2",
      "speakingRate": 1.0,
      "pitch": 0.0
    }
  }
}

Or keep using ElevenLabs by not setting ttsProvider (or setting it to "elevenlabs").

Files Changed

• Releases/v3.0/.claude/VoiceServer/server.ts — Multi-provider TTS routing, Google Cloud TTS implementation

Context

This is a re-implementation of PR #285 (merged 2026-01-01, lost in v3.0 restructuring) targeting the current v3.0 architecture. The original code lived in Packs/kai-voice-system/ which no longer exists.

Closes #682

[kaimagnus]

Nice addition of Google Cloud TTS! This has merge conflicts with recent VoiceServer changes (commit 95d65cc). Could you rebase on main? We'll merge once the conflicts are resolved. Thanks! 🙏

[fayerman-source]

Rebased! The conflict was in loadVoiceConfig's return statements — your desktopNotifications field and our new ttsProvider/googleVoice fields both needed to land in LoadedVoiceConfig, so I merged them all in. Three return sites updated. Should be clean now. Ready when you are!

[danielmiessler]

Thank you for adding Google Cloud TTS as an alternative provider! This PR targets the v3.0 VoiceServer architecture, which was restructured in v4.0.

If you'd like to contribute alternative TTS provider support for the current architecture, please see Releases/v4.0.2 for the latest codebase and feel free to resubmit. Thanks for the contribution!