FireRedTeam
Open-source speech models from Xiaohongshu's Audio Lab — ASR, TTS, and real-time voice chat.
About FireRedTeam
FireRedTeam is the open-source speech-model effort from Xiaohongshu's Super Intelligence Audio Lab. It releases production-grade models across the full speech stack — automatic speech recognition (FireRedASR), text-to-speech (FireRedTTS2), and real-time voice chat (FireRedChat) — freely available on GitHub for developers building voice AI.
Open-source projects
- FireRedASR — high-accuracy automatic speech recognition (STT)
- FireRedTTS2 — natural, expressive text-to-speech generation
- FireRedChat — real-time, full-duplex conversational voice agent
Who Is It For?
- Developers and researchers building speech and voice-agent applications
- Teams wanting open, self-hostable ASR / TTS / voice-chat models
Use Cases
- Transcription and speech understanding with FireRedASR
- Voice generation with FireRedTTS2
- Real-time voice assistants and chat with FireRedChat
Explore FireRedTeam's open-source speech models on GitHub.