AI voice company Fish Audio has released its next-generation voice generation model, Drama
3, which focuses on improving AI voice's ability to express emotions, control tone, and reproduce details. The company stated that the new model aims to allow developers to fine-tune pauses, rhythm, and expression in voice to meet the needs of scenarios such as character dubbing, AI assistants, and digital humans.
In recent years, Fish Audio has primarily focused on text-to-speech (TTS) and voice cloning technologies. Its S2 series models already support voice performance control via natural language commands. As AI applications evolve from text interaction to voice interaction, the naturalness, controllability, and real-time generation capabilities of voice models have become key areas of competition in the industry.