# AI use cases for Openspline (openspline/openspline) Repo: [openspline/openspline](https://github.com/openspline/openspline) | Use cases: 4 | Page: https://dedupedai.com/use-cases/openspline-openspline/ OpenSpline is a Python and Node SDK that turns a portrait and generated audio into a synchronized talking avatar video, on OpenSpline Cloud or a self-hosted NVIDIA GPU. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Add a live talking face to a Python voice agent: https://dedupedai.com/use-cases/openspline-openspline/add-live-face-to-python-voice-agent/ - Run a local avatar demo on your own NVIDIA GPU: https://dedupedai.com/use-cases/openspline-openspline/run-avatar-demo-on-own-nvidia-gpu/ - Put a face on an OpenAI Realtime agent: https://dedupedai.com/use-cases/openspline-openspline/put-face-on-openai-realtime-agent/ - Add an avatar to a LiveKit voice agent: https://dedupedai.com/use-cases/openspline-openspline/add-avatar-to-livekit-agent/ ## How to add a live talking face to a Python voice agent with Openspline Repo: [openspline/openspline](https://github.com/openspline/openspline) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Stream your agent's generated audio into a portrait to show a synced avatar. **Who it is for:** A developer with a working Python voice agent who wants a visible face for demos or customer-facing sessions. **The problem:** Voice agents are audio-only, so users have no face to read or connect with during a conversation. **The outcome:** A viewer URL that shows a portrait speaking the agent's audio in sync. ### Steps 1. Run `pip install openspline`. 2. Set `OPENSPLINE_API_KEY` on your backend, or run a local service as described in the Self Host section. 3. Create the client and open the avatar with `async with client.avatar("portrait.jpg", quality="low") as avatar:`, then print `avatar.viewer_url`. 4. Call `await avatar.stream(agent.audio_stream(), sample_rate=24000)` with your agent's PCM audio. 5. Open the viewer URL in a browser while the agent speaks. ### Prompt to paste into your agent ```text Use the openspline repo (github.com/openspline/openspline) to add an avatar to the Python voice agent at . Use portrait and pass the agent's PCM audio to avatar.stream with sample_rate=24000. Output the viewer URL and confirm the agent's audio plays once, not twice. ``` ### Caveat The README says the GPU stays active during a connected session and smooth motion requires inference to keep up with playback. Risk, credentials: Keep the OpenSpline API key and any provider credentials on your backend, not in browser code. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/openspline-openspline/add-live-face-to-python-voice-agent/ ## How to run a local avatar demo on your own NVIDIA GPU with Openspline Repo: [openspline/openspline](https://github.com/openspline/openspline) | Difficulty: Intermediate | Time: 60 minutes | Basis: From the README Install the GPU service and open a local demo at localhost:7860. **Who it is for:** A developer with a Linux or WSL2 machine and an NVIDIA GPU who wants to avoid cloud costs and keep generation local. **The problem:** Cloud avatar generation may be unavailable or unwanted, so the developer needs to run the models locally. **The outcome:** A working local demo at http://localhost:7860 that generates avatar video on the local GPU. ### Steps 1. Confirm you are on a Linux device or WSL2 with an NVIDIA GPU. AMD and Apple GPUs are not supported. 2. Run `curl -fsSL https://raw.githubusercontent.com/openspline/openspline/main/install.sh | sh`. 3. Run `sh ~/.local/share/openspline/run.sh`. 4. Open http://localhost:7860 in a browser. 5. Check the GPU service health at `/readyz` on the local address before connecting an SDK client. (inferred, check the README) ### Prompt to paste into your agent ```text Use the openspline repo (github.com/openspline/openspline) to set up the local GPU service on . Run the install script and launch the demo, then confirm the demo answers at http://localhost:7860. Report the GPU model and whether the demo page loads. ``` ### Caveat The README says Blackwell and pre-Ampere inference still need hardware validation, and pre-Ampere cards use FP32 with more VRAM needed. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/openspline-openspline/run-avatar-demo-on-own-nvidia-gpu/ ## How to put a face on an OpenAI Realtime agent with Openspline Repo: [openspline/openspline](https://github.com/openspline/openspline) | Difficulty: Advanced | Time: 60 minutes | Basis: From the README Wrap an existing OpenAI Realtime connection so the avatar speaks its replies. **Who it is for:** A developer whose agent already uses the OpenAI Realtime API and needs a visual avatar without rewriting event handling. **The problem:** A Realtime agent plays audio only, and wiring audio to a video avatar by hand is error prone. **The outcome:** The existing Realtime event handler keeps working while the avatar speaks the agent's audio. ### Steps 1. Run `pip install openai==2.54.0`. 2. Set `OPENAI_API_KEY` in your backend environment. 3. Open the avatar with `async with Openspline().avatar("portrait.jpg") as avatar:` and wait with `await avatar.wait_for_viewer()`. 4. Create `adapter = OpenAIRealtime(avatar, connection)` and loop over `adapter.wrap(connection)`, passing each event to your handler. 5. Disable your previous speaker output so audio does not play twice, then run `OPENAI_API_KEY=... python examples/python/openai_realtime.py portrait.jpg`. ### Prompt to paste into your agent ```text Use the openspline repo (github.com/openspline/openspline) to add the OpenAIRealtime adapter to the agent in , keeping its existing event handler. Use portrait . Confirm the speaker output is disabled so audio plays once, and report the viewer URL. ``` ### Caveat The README pins openai==2.54.0 for this connector, so newer SDK versions may not match the example. Risk, credentials: Keep OPENAI_API_KEY on the backend and never expose it to the browser. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/openspline-openspline/put-face-on-openai-realtime-agent/ ## How to add an avatar to a LiveKit voice agent with Openspline Repo: [openspline/openspline](https://github.com/openspline/openspline) | Difficulty: Advanced | Time: 90 minutes | Basis: From the README Attach an avatar to your AgentSession so it plays in your normal LiveKit frontend. **Who it is for:** A team running a LiveKit voice agent that needs a face shown inside its existing room frontend. **The problem:** LiveKit rooms carry audio only, so the agent has no visible presence for participants. **The outcome:** The avatar appears in the same LiveKit room as the agent, driven by its audio. ### Steps 1. Run `pip install 'livekit-agents[openai]==1.8.3'`. 2. Set `OPENAI_API_KEY`, `LIVEKIT_URL`, `LIVEKIT_API_KEY`, and `LIVEKIT_API_SECRET`. 3. Create `avatar = AvatarSession("portrait.jpg", client=Openspline(quality="low"))` and call `await avatar.start(agent_session, room=ctx.room)` before your existing session starts. 4. Run `PORTRAIT=portrait.jpg python examples/python/livekit_agent.py dev`. 5. Join the room from your normal LiveKit frontend and confirm the avatar speaks. ### Prompt to paste into your agent ```text Use the openspline repo (github.com/openspline/openspline) to attach an AvatarSession to the LiveKit agent in , using portrait . Start the avatar before agent_session.start and confirm the avatar is visible in the room. ``` ### Caveat The README says the native service includes LiveKit support, so check that your setup uses that service before relying on this connector. Risk, credentials: LiveKit API secret and OpenAI key must stay on the server side of the agent. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/openspline-openspline/add-avatar-to-livekit-agent/