[Suggestion] Google Genai Processors Deepgram STT provider plugin (Python)

[deepgram-robot]

What to build

A Deepgram STT processor plugin for Google's genai-processors pipeline framework, enabling developers to compose Deepgram transcription into Google AI pipelines using the operator-overloaded stt + llm + tts syntax.

Why this matters

Google's genai-processors framework provides a composable pipeline architecture for building multimodal AI applications. Developers using this framework browse the available processors to select their preferred STT, LLM, and TTS providers. Without a Deepgram processor, developers building pipelines in this ecosystem default to other STT providers. A contrib-style processor makes Deepgram discoverable in a growing AI pipeline ecosystem and lets developers drop Nova-3 into any genai-processors pipeline with a single import.

Suggested scope

• Language: Python
• Framework: google/genai-processors
• Deepgram APIs: Streaming STT (WebSocket), pre-recorded STT (REST)
• What to include:
  • A DeepgramSTTProcessor class implementing the genai-processors Processor interface
  • Support for both streaming and batch transcription modes
  • Configuration for model selection (nova-3, flux), language, keywords, and other STT parameters
  • A complete pipeline example: microphone → Deepgram STT → Gemini LLM → console output
  • Unit tests and README with setup instructions

Acceptance criteria

• Runnable with minimal setup (clone, add API key, run)
• README explains the pattern clearly
• Uses current SDK version
• Processor composes correctly with other genai-processors using + and // operators
• Includes both streaming and batch transcription modes
• Works with the genai-processors visual tracing output

---

Raised by the DX intelligence system.