Sources
Fish Audio S2: Open-Source TTS with Multi-Speaker, Multi-Turn, and Instruction-Following Control
Trending on HuggingFace Daily Papers, Fish Audio S2 is an open-source text-to-speech system that adds natural-language instruction-following control over voice characteristics, enabling multi-speaker and multi-turn generation. Significant for builders adding voice interfaces to agentic systems—instruction-following TTS eliminates the need for voice-specific fine-tuning pipelines and aligns TTS configuration with the same prompt-based paradigm used for LLM orchestration.
↳ Follow the thread