/
#101
Fish Audio
Voice cloning and TTS platform competing with ElevenLabs. Their open-source Fish Speech S2 Pro model uses a Dual-Autoregressive architecture on a Qwen3 backbone, trained on 10M+ hours across 80+ languages with GRPO reinforcement learning alignment. On Seed-TTS Eval, S2 posts the lowest WER of any evaluated system, open or closed source.
Links