🎤 Whisper Speaker Diarization Demo

AI-powered speaker identification and transcription

Telecom Paris

🔬 Processing Information

  • 💻 CPU processing only (slower than GPU)
  • 📦 No file size limits - process any audio length
  • 🌍 For multi-language audio, use larger models (medium, large-v2, large-v3)
  • ⚡ Larger models provide better accuracy but take longer to process
  • ⚠️ Very large files may take significant time and memory

Supported: MP3, WAV, M4A, FLAC, etc.
🎯 Whisper Model
🌍 Language
🎯 Processing Mode

Standard: Traditional diarization | Separation: Pre-separate speakers
1 4

📚 How to Use

  1. Upload audio (any size)
  2. Choose processing mode
  3. Configure settings (optional)
  4. Click process and wait
  5. Download results

🎯 Processing Modes

  • Standard: Traditional speaker diarization
  • Speaker Separation: Pre-separate speakers first

🌍 Model Selection

  • tiny.en/base.en/small.en: Fast, English only
  • medium.en: Better accuracy, English only
  • medium/large-v2/large-v3: Best for multi-language audio

⚠️ Large File Warning

  • Large files will take longer to process
  • Monitor system resources during processing