Fish Audio is an AI voice generation platform built by Hanabi AI Inc. that has become one of the most popular alternatives to ElevenLabs, powered by its OpenAudio S1 and flagship S2 Pro text-to-speech models — top-ranked performers in 2026 blind TTS evaluations. The platform generates ultra-low-latency, emotionally controllable speech using inline emotion tags (angry, sad, excited, whispering), clones voices from roughly 15 seconds of reference audio, and supports more than 30 languages including English, Japanese, Korean, Chinese, French, German, Arabic, and Spanish. Beyond core TTS, Fish Audio bundles speech-to-text with multi-speaker and emotion-tag support, a Voice Agent stack for real-time conversational applications, and Story Studio for audiobook production. Developers can integrate the same models through a pay-as-you-go REST API (the s2-pro model is priced at $15 per million UTF-8 bytes), while consumer plans scale from a free tier to team-oriented Pro and Max subscriptions with commercial usage rights.