feat: add missing ElevenLabs models and voices

[Kamilbenkirane]

Changes

New Models

Added four missing ElevenLabs models:

• eleven_v3 - Eleven v3 model
• eleven_turbo_v2_5 - Eleven Turbo v2.5 model
• eleven_flash_v2 - Eleven Flash v2 model
• eleven_multilingual_v1 - Eleven Multilingual v1 model

All models support:

• Voice selection from available voices
• Speed control (0.7-1.2)
• Multiple response formats (mp3_44100_128, pcm_22050_16, pcm_24000_16, pcm_44100_16)
• Streaming support

New Voice

• Added Spuds voice (Grandpa Spuds Oxley) - ID: NOpBlnGInO9m6vDvFkFC

Version Bumps

• celeste-speech-generation: 0.2.8 → 0.2.9
• celeste-ai: 0.2.13 → 0.2.14
• Updated dependency requirements in root pyproject.toml

Testing

• All models follow the same parameter constraints pattern as existing ElevenLabs models
• Voice constraint updated to include new Spuds voice

[claude]

Code Review: PR #59 - Add missing ElevenLabs models and voices

Summary

This PR adds four new ElevenLabs speech generation models and one new voice. The changes are clean, well-structured, and follow existing patterns in the codebase.

✅ Code Quality & Best Practices

Excellent:

• ✅ Consistent code structure - all new models follow the exact same pattern as existing models
• ✅ Proper typing and structure with Model, VoiceConstraint, Range, and Choice
• ✅ Clear, descriptive display names
• ✅ Appropriate version bumping (minor version increments)
• ✅ Good inline documentation for the new voice
• ✅ Clean, readable code with proper formatting

Observations:

• The models have significant code duplication (all 4 new models + existing ones share identical parameter constraints). This is acceptable given the declarative nature of the configuration, but consider extracting common constraints if this list grows significantly larger.

🔍 Potential Issues

Model Order (Minor)

The models list is not in a clear ordering pattern. Consider organizing by:

• Version (newest to oldest), OR
• Alphabetically, OR
• By use case (turbo vs multilingual vs flash vs standard)

Current order appears somewhat mixed. For maintainability, I'd suggest ordering newest to oldest within each model family:

eleven_v3
eleven_turbo_v2_5
eleven_turbo_v2
eleven_flash_v2_5
eleven_flash_v2
eleven_multilingual_v2
eleven_multilingual_v1
eleven_monolingual_v1

This helps developers quickly find the latest models to use.

Verify Model IDs

Ensure that model IDs match ElevenLabs' official documentation:

• ✅ eleven_v3
• ✅ eleven_turbo_v2_5
• ⚠️ eleven_flash_v2 - Please verify this exists (I see eleven_flash_v2_5 already in the codebase)
• ✅ eleven_multilingual_v1

These should be confirmed against the ElevenLabs API documentation.

Voice ID Verification

• ⚠️ Voice ID NOpBlnGInO9m6vDvFkFC for "Spuds" - please confirm this is correct from the ElevenLabs API, as voice IDs must be exact

⚡ Performance Considerations

No concerns:

• These are declarative configuration additions with no runtime performance impact
• All models support streaming, which is good for performance
• Speed constraints (0.7-1.2) are reasonable and consistent

🔒 Security Concerns

No security issues identified:

• No user input handling
• No API key exposure
• Configuration-only changes
• Proper constraint validation in place

🧪 Test Coverage

Needs attention:

The existing test file packages/speech-generation/tests/integration_tests/test_speech_generation/test_generate.py:19 currently tests eleven_flash_v2_5.

Recommendations:

1. Add parametrized tests for the new models to ensure they work correctly:

   • eleven_v3
   • eleven_turbo_v2_5
   • eleven_flash_v2
   • eleven_multilingual_v1

2. Test the new voice - add a test case using the "Spuds" voice to verify it works

3. Consider adding a model registry test that validates:

   • All models have unique IDs
   • All models have required parameter constraints
   • Model IDs match expected patterns

Example test to add:

@pytest.mark.parametrize(
    ("model", "voice"),
    [
        ("eleven_v3", "Spuds"),
        ("eleven_turbo_v2_5", "Laura"),
        ("eleven_flash_v2", "Adam"),
        ("eleven_multilingual_v1", "Bella"),
    ],
)
@pytest.mark.integration
@pytest.mark.asyncio
async def test_new_elevenlabs_models(model: str, voice: str) -> None:
    """Test newly added ElevenLabs models."""
    # ... test implementation

📋 Additional Recommendations

1. Documentation: Consider updating the README or docs to mention the new models if they have specific use cases (e.g., "Use eleven_v3 for highest quality", "Use eleven_flash_v2 for lowest latency")

2. Deprecation: If any older models should be deprecated in favor of these new ones, add deprecation warnings

3. Changelog: Ensure the changes are documented in a CHANGELOG.md if the project maintains one

Verdict

Recommendation: Approve with minor suggestions

The code is solid and follows best practices. The main action items are:

1. ✅ Verify model IDs against ElevenLabs API docs
2. ✅ Verify the Spuds voice ID
3. 🧪 Add test coverage for new models
4. 📝 Consider reordering models for better maintainability

Great work on maintaining consistency with the existing codebase! 🎉

---

Review generated by Claude Code