Push `TranscriptionFrame`s upstream, to be handled by the user context aggregator. This will require at least a couple of other changes:
- Updating examples to put transcript processors upstream from `OpenAIRealtimeLLMService`
- Maybe figuring out a way to preserve backward compatibility with existing pipelines that put transcript processors downstream from `OpenAIRealtimeLLMService`
- Updating `OpenAIRealtimeLLMService` to ignore new received context frames, since the upstream user context aggregator will generate those after each newly-added user message; hopefully nobody was reliant on the old behavior of resetting the conversation upon receiving a new context!
Avoid pushing `LLMTextFrame` when `OpenAIRealtimeLLMService` is configured to output audio. This avoids duplicate text in assistant messages in context. Conceptually, a speech-to-speech service encapsulates TTS behavior; in a "traditional" pipeline, `LLMTextFrames` are swallowed by the TTS service, so they should similarly not be pushed by a speech-to-speech service. Only. `TTSTextFrame`s should be pushed.
Properly bookended responses now work with:
- AUDIO modality (validated with 26b example)
- TEXT modality (validated with 26d example)
- AUDIO modality with Vertex AI (validated with 26h example)
It doesn't seem that TEXT modality is supported with Vertex AI, hence the missing "quadrant" of validation.
- Change GenerationConfig from dataclass to Pydantic BaseModel for consistency
- Simplify _build_msg() to use model_dump(exclude_none=True) instead of manual field extraction
- Simplify HTTP run_tts() to use model_dump(exclude_none=True) instead of manual field extraction
This addresses feedback from code review and reduces code duplication.
Add GenerationConfig dataclass with volume, speed, and emotion parameters
for Cartesia Sonic-3 TTS models. This enables fine-grained control over
speech generation including volume (0.5-2.0), speed (0.6-1.5), and
emotion (60+ options).
Changes:
- Add GenerationConfig dataclass with proper Google-style docstrings
- Update CartesiaTTSService.InputParams to include generation_config
- Update CartesiaHttpTTSService.InputParams to include generation_config
- Modify _build_msg() to include generation_config in WebSocket messages
- Modify run_tts() to include generation_config in HTTP requests
- Maintain backward compatibility with existing speed and emotion parameters
The legacy speed (literal strings) and emotion (list) parameters remain
available for non-Sonic-3 models.
This wasn't really an issue before, when folks were *knowingly* migrating from `OpenAILLMContext` to `LLMContext`. But in the latest AWS Nova Sonic change, we're swapping it out from under folks, so this kind of compatibility is more important.
For context, the reason we *didn't* offer the `messages` property earlier was to aid in the development of `LLMContext`—we wanted to draw attention to all the places where messages were being read from context, so we could find the places where we might need to pass an argument to the read.
