Add a face to your NVIDIA Riva pipeline

LemonSlice puts a lifelike, interactive avatar on top of your NVIDIA speech pipeline. Whether you run Riva TTS, Riva STT, or NIM microservices, the avatar turns that audio into natural, face-to-face conversation.

LemonSlice

At a glance

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

How the integration works

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

  1. Keep your NVIDIA setup

    Your Riva TTS and STT services and NIM microservices keep running on your own GPU infrastructure. Nothing about your pipeline, models, or deployment changes.

  2. Stream the audio to LemonSlice

    Send the audio your pipeline generates to the LemonSlice avatar API. The avatar renders in real time with accurate lip sync and expression.

  3. The avatar video streams to your users

    LemonSlice delivers the avatar video over WebRTC, natively via LiveKit, Daily (Pipecat), or Agora, so it drops into your existing transport.

Ship your interactive avatar with simple code

A NVIDIA 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 NVIDIA for TTS and STT, plus your LLM, 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 lifelike avatar in every conversation

Control actions and emotions

The Action Engine animates everything from the face to hands, shoulders, and posture. You trigger actions and emotions from the API, down to a prompt like "talk with hand gestures."

Works with any character, in any style

LemonSlice is the only interactive avatar model able to animate non-humans, a turtle or a smiling teacup, with the same ease as cartoons and photorealistic people.

Ultra-low latency

LemonSlice offers the fastest responses of any major avatar provider, so replies feel instant.

Create an avatar from one image

A single photo is enough to create an interactive avatar. No training or fine-tuning needed. Make unlimited custom characters and change outfits or settings in real time.

Built for enterprise scale

Avatar sessions alongside NVIDIA run at production scale.

Thousands of concurrent calls

Run thousands of concurrent avatar sessions, ready for high-volume production workloads.

Multi-region servers

Choose the region closest to your users for avatar sessions so latency stays minimal.

Zero data retention mode

Run entirely without stored conversation data, made for privacy-sensitive enterprise workflows.

What builders ship with NVIDIA + LemonSlice

Create avatars for everything from customer support to corporate training, airport kiosks, and language practice.

Customer support agent

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

Corporate training coach

Roleplay difficult conversations and train staff on policy.

Airport help desk

Wayfinding and flight answers on a kiosk screen.

Frequently asked questions

Stream the audio your Riva pipeline produces to the LemonSlice avatar API, and the avatar video streams back to your users over WebRTC via LiveKit, Daily (Pipecat), or Agora.
No. Riva handles speech recognition and synthesis in your pipeline, and LemonSlice renders the interactive avatar that speaks the output.
Yes. Riva running on your own GPU infrastructure streams audio to the LemonSlice API the same way a hosted service would, with no changes to your deployment.
Yes. Any NIM-based conversational pipeline that produces audio can stream that audio to LemonSlice for real-time avatar rendering.
Riva TTS, Riva STT, and NIM microservices all work, since the interactive avatar speaks whatever audio your pipeline generates.
No. Your Riva services, NIM microservices, models, and GPU infrastructure stay exactly as they are, and the avatar reads from the audio your pipeline already sends.

Give your NVIDIA pipeline a face

5-minute API setup, and your existing Riva stack stays untouched.