Nara Lens/Docs

Introduction

Welcome to Nara Lens, the API for Active Speaker Detection in video.

What is Nara Lens?

Nara Lens provides an API for detecting active speakers in video content. Our system analyzes video frames to identify faces, track individuals, and determine who is speaking at any given moment.

Key Features

  • Active Speaker Detection - Identify who is speaking in each frame
  • Face Detection - Detect and track faces throughout the video
  • Speaking Scores - Get confidence scores for speaker activity
  • Person Tracking - Track individuals across frames with unique IDs
  • Async Processing - Submit videos and poll for results

How It Works

  1. Submit - POST your video URL to /api/vad/analyze
  2. Process - Our CV service analyzes the video
  3. Poll - GET results from /api/vad/result/{task_id}
  4. Use - Parse frame-by-frame speaker detection data

Quick Example

Bash
# Submit video for analysis
curl -X POST https://naralens.com/api/vad/analyze \
  -H "Authorization: Bearer $NARA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"video_url": "https://example.com/video.mp4"}'

# Response: {"task_id": "abc123", "status": "pending"}

Next Steps