> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/nim/maxine/llms.txt.

# Maxine NIM Overview

NVIDIA Maxine NIMs are a collection of GPU-accelerated inference microservices that bring AI-powered video and audio effects behind consistent service interfaces. Each NIM packages a model as a self-hosted, easily deployable container, so you can add effects such as eye contact, relighting, lip sync, speech enhancement and noise removal to live and recorded media workflows. Holoscan for Media variants deploy the same capabilities into live production environments over SMPTE ST 2110.

## Use Cases

Maxine NIMs span video enhancement, video analysis, speech-driven animation, audio enhancement and live production.

### Video Enhancement

Use these NIMs when you need to improve or restyle video:

<CardGroup cols={3}>
  <Card title="Eye Contact" href="/eye-contact">
    Redirects a user's eye position toward the camera to simulate natural eye contact.
  </Card>
  <Card title="Relighting" href="/relighting">
    Applies lighting effects to videos using HDR environment maps.
  </Card>
  <Card title="Video Super Resolution" href="/vsr">
    Upscales video to higher resolutions for ST 2110 and compressed-video workflows.
  </Card>
</CardGroup>

**Benefits**: Self-hosted GPU-accelerated containers, gRPC service interfaces

### Video Analysis

Use these NIMs when you need to understand what is in a video:

<CardGroup cols={3}>
  <Card title="3D Body Pose" href="/body-pose">
    Estimates the 3D pose of every tracked person in a video.
  </Card>
  <Card title="Synthetic Video Detector" href="/synthetic-video-detector">
    Estimates the likelihood that a video has been AI-generated.
  </Card>
  <Card title="Active Speaker Detection" href="/active-speaker-detection">
    Detects and identifies active speakers in a video stream from visual and diarized audio data.
  </Card>
</CardGroup>

**Benefits**: Per-frame and per-stream analysis, self-hosted containers, gRPC service interfaces

### Speech-Driven Animation

Use these NIMs when you need facial movement that follows speech:

<CardGroup cols={2}>
  <Card title="LipSync" href="/lipsync">
    Synchronizes lip movements in videos with input speech.
  </Card>
  <Card title="Audio2Face-2D" href="/audio2face-2d">
    Creates facial animations from a portrait photo and driving audio.
  </Card>
</CardGroup>

**Benefits**: Speech audio combined with video or a portrait photo, self-hosted containers

### Audio Enhancement

Use these NIMs when you need cleaner speech audio:

<CardGroup cols={2}>
  <Card title="Studio Voice" href="/studio-voice">
    Enhances speech recorded through low-quality microphones to studio-recorded quality.
  </Card>
  <Card title="Background Noise Removal" href="/bnr">
    Removes unwanted background noise from audio.
  </Card>
</CardGroup>

**Benefits**: Self-hosted GPU-accelerated containers, gRPC service interfaces

### Holoscan for Media

Use these NIMs when you need Maxine effects in live production over SMPTE ST 2110:

<CardGroup cols={2}>
  <Card title="Studio Voice on Holoscan for Media" href="/studio-voice-h4m">
    Studio Voice speech enhancement for Holoscan for Media deployments.
  </Card>
  <Card title="Background Noise Removal on Holoscan for Media" href="/bnr-h4m">
    Background noise removal for a single PCM audio stream over SMPTE ST 2110-30.
  </Card>
  <Card title="LipSync on Holoscan for Media" href="/lipsync-h4m">
    LipSync for Holoscan for Media deployments.
  </Card>
  <Card title="Active Speaker Detection on Holoscan for Media" href="/active-speaker-detection-h4m">
    Active speaker detection in live video streams.
  </Card>
</CardGroup>

**Benefits**: Helm chart and Kubernetes operator deployment, SMPTE ST 2110 and NMOS integration