Boson AI logo

Higgs Avatar

A face for your voice agent, live in the call.

Connect Higgs Avatar to the real-time agent you already run. It joins the call as a face that speaks your agent's words.

Try Higgs AvatarGet an API key

Talk to any face, rendered from one image.

Every face below is Higgs Avatar in a different role, and none of them needed custom training. Pick a card to watch a recorded call.

  • Coaching

  • Interview

  • Cartoon

  • Insurance claims

How it works.

  • Streams every frame live

    We took a pre-trained video model and adapted it to generate one frame at a time. Each frame streams alongside the audio as the conversation happens.

  • Starts from one still image

    A single photo is all it needs. Co-designed with Higgs TTS so lip-sync, expression, and head motion stay locked to the voice.

  • Runs eight conversations per GPU

    A single H100 hosts eight conversations at once, which keeps the per-conversation cost low enough for production.

  • Compatible with Higgs Realtime and Higgs TTS

    Give the face a Higgs voice: Higgs Realtime for live conversations, Higgs TTS for scripted speech.

    Higgs RealtimeHiggs TTS

Renders faster than real time.

Higgs Avatar generates each frame in about 16 ms. A live conversation needs a new one every 62.5 ms, so the face never falls behind the voice.

Priced for production.

Pay only for the minutes of video Higgs Avatar produces, live in a call or rendered to a file. The audio and text you send in are free.

Full pricing

Output video length

Price

640 × 480
$0.07 / min
480 × 640
$0.07 / min
640 × 640
$0.09 / min
Audio + text input
Free

Live calls and rendered video are billed at the same per-minute rates.

Add it to your LiveKit agent.

A few lines of code connect Higgs Avatar to the LiveKit agent you already run, and the face joins every call. Need a video file instead? Send a script and get the video back.

Read API docs
from livekit.agents import AgentSession
+from livekit.plugins import boson
 
async def entrypoint(ctx):
    session = AgentSession(stt=..., llm=..., tts=...)  # yours
+   avatar = boson.AvatarSession(face="camila")
+   await avatar.start(session, room=ctx.room)
    await session.start(room=ctx.room, agent=Assistant())

Put a face on your agent's next call.

Higgs Avatar is available now via API, for live calls and rendered video. Get an API key to start, or try Higgs Avatar in Workspace.

Try Higgs AvatarGet an API key