Skip to main content
Boson AI provides infrastructure for real-time voice and avatar experiences. The Higgs API gives developers a family of models for speech generation, audio understanding, lifelike avatars, and interactive agents.

​
Higgs models

Higgs Audio v3 TTS

Natural, expressive text-to-speech in 100 languages, with controllable style for real-time voice applications.

Higgs Avatar v1

Avatar video generation from a single still image and a driving voice, built for real-time interaction.

Higgs Audio v3 Instruct

Audio-in LLM with voice-agent reflexes: understands speech and text, follows instructions, calls tools, handles interruptions, and returns grounded text output.