When will Amazon Nova Sonic support the half-cascased architecture and accept a custom TTS for the voice part #252
Description
I'm using Amazon Nova Sonic 2 to build a Voice AI Agent using Livekit. I would like to use the half-cascaded architecture and use a custom TTS (from Eleven Labs or Cartesia) instead of using the available voices from Amazon.
Popular speech-to-speech models like OpenAI & Gemini Live already support the half-cascaded architecture. I would like Amazon Nova Sonic to also supporting this as that's important for the use case I'm working on.
Is this something that's on the roadmap for Amazon Nova Sonic??