Add a face to your AssemblyAI pipeline

LemonSlice puts a lifelike, real-time avatar on top of AssemblyAI's speech stack. While AssemblyAI transcribes your callers with streaming speech-to-text, the avatar turns your agent's audio into natural, face-to-face conversation.

LemonSlice

At a glance

Category
Transcription (STT + VAD)
Status
Compatible (works with any)
Type
Avatar API
Languages
Python, Node.js
Setup time
~5 minutes

How the integration works

AssemblyAI keeps doing what it does today. LemonSlice turns the audio into a face.

  1. Keep your AssemblyAI setup

    AssemblyAI keeps handling real-time transcription of your users' speech. Your models, endpoints, and pipeline logic stay exactly as they are. STT choices never change how the avatar renders.

  2. Stream the audio to LemonSlice

    When your agent responds, send its speech audio to the LemonSlice avatar API. The avatar renders in real time with lip sync and expression matched to the audio.

  3. The avatar video streams to your users

    LemonSlice delivers the avatar video over WebRTC, natively via LiveKit, Daily (Pipecat), or Agora. Your users see a responsive face on the other end of the call.

Add your avatar with one code block

A AssemblyAI session with a LemonSlice avatar.

Choose a code example
from livekit import agents
from livekit.agents import AgentSession, RoomOutputOptions
from livekit.plugins import lemonslice

async def entrypoint(ctx: agents.JobContext):
    await ctx.connect()

    session = AgentSession(
        # Add AssemblyAI STT, plus your LLM and TTS, here
    )

    avatar = lemonslice.AvatarSession(
        # Publicly accessible image URL for the avatar
        agent_image_url="...",
        # Prompt to guide the avatar's demeanor while speaking
        agent_prompt="a person talking",
        # Recommended. Adapts video quality to each viewer's bandwidth
        simulcast=True,
    )

    # Start the avatar and wait for it to join
    session_id = await avatar.start(session, room=ctx.room)

    # Start your agent session with the user
    await session.start(
        # ... room, agent, room_options, etc....
    )

A real-time avatar that feels human

Control actions and emotions

The Action Engine moves face, hands, shoulders, and posture together. You control actions and emotions through the API, down to a prompt like "talk with hand gestures."

Works with any character, in any style

LemonSlice is the only avatar model that brings non-humans to life (a smiling teacup, a turtle) as easily as photorealistic people and cartoons.

Ultra-low latency

LemonSlice leads every major avatar provider on response time, keeping conversations feeling instant.

Create an avatar from one image

Turn any image into an interactive avatar. No training or fine-tuning required. Create unlimited custom characters with real-time outfit or setting changes.

Built for enterprise scale

Avatar sessions alongside AssemblyAI run at production scale.

Thousands of concurrent calls

Scale avatar sessions into the thousands, designed for high-volume production workloads.

Multi-region servers

Place avatar sessions in the region nearest your users to keep latency low.

Zero data retention mode

Store no conversation data during avatar sessions, designed for privacy-sensitive enterprise workflows.

What builders ship with AssemblyAI + LemonSlice

Create avatars for everything from language practice to customer support, live streaming, and corporate training.

Language practice partner

Practice real conversations in another language.

Customer support agent

Answer, triage, and resolve customer questions face to face.

Streaming host

A live host that reads and responds to chat.

Frequently asked questions

Keep AssemblyAI transcribing user speech, then stream your agent's response audio to the LemonSlice avatar API. The real-time avatar renders with lip sync and expression, and the video streams back over WebRTC via LiveKit, Daily (Pipecat), or Agora.
No. AssemblyAI handles the input side by transcribing what your users say, while LemonSlice handles the output side by rendering the real-time avatar that speaks back.
No. The avatar animates from the response audio it receives, so AssemblyAI's transcription layer stays fully independent of how the avatar looks and moves.
No. AssemblyAI keeps receiving the same user audio it does today, so transcription accuracy is unaffected. LemonSlice only consumes your agent's outgoing speech.
Any AssemblyAI setup that powers a voice agent works, including its streaming speech-to-text built for real-time transcription. LemonSlice only needs the response audio your agent produces.
No. LemonSlice is built for live conversation and delivers the fastest response time of any major avatar provider, so the exchange feels instant rather than delayed.

Give your AssemblyAI agent a face

Keep your transcription pipeline, add a real-time avatar with a 5-minute API setup.