jonesckevin/whisper-webapp repository overview

⁠Whisper Transcription Tool

DockerHub - Jonesckevin⁠

AI-powered audio/video transcription using OpenAI Whisper with GPU acceleration. And it's a WebApp in Docker for all your HALp needs!

⁠Quick Start

Choose Your Version:

• Local CLI: .\run_whisper.ps1 (Windows) or python src/audio_to_text4.py input.mp4
• Docker Web: .\start-docker.ps1 → option 1 → http://localhost:8080⁠
• Docker CLI: .\start-docker.ps1 → option 2 (batch processing)
• DockerHub:

# CPU Version
docker run -d \
  --name whisper-webapp \
  -p 8000:5000 \
  -v "$(pwd)/data/uploads:/data/uploads" \
  -v "$(pwd)/data:/data/db" \
  -v "$(pwd)/data/models:/root/.cache/whisper" \
  -e NVIDIA_VISIBLE_DEVICES=all \
  -e CUDA_VISIBLE_DEVICES=0 \
  -e FLASK_ENV=production \
  -e MAX_UPLOAD_SIZE_GB=5 \
  -e PRELOAD_WHISPER_MODELS=false \
  --gpus '"device=0"' \
  --restart unless-stopped \
  jonesckevin/whisper-webapp:gpu

# GPU Version
docker run -d \
  --name whisper-webapp \
  -p 8000:5000 \
  -v "$(pwd)/data/uploads:/data/uploads" \
  -v "$(pwd)/data:/data/db" \
  -v "$(pwd)/data/models:/root/.cache/whisper" \
  -e NVIDIA_VISIBLE_DEVICES=all \
  -e CUDA_VISIBLE_DEVICES=0 \
  -e FLASK_ENV=production \
  -e MAX_UPLOAD_SIZE_GB=5 \
  -e PRELOAD_WHISPER_MODELS=false \
  --gpus '"device=0"' \
  --restart unless-stopped \
  jonesckevin/whisper-webapp:gpu

⁠Docker Web Features

Access: http://localhost:8080⁠ after running .\start-docker.ps1 → option 1

• 📤 Upload: Drag & drop files (max 5GB via web, unlimited via network share)
• 📋 Queue: View pending jobs, reorder priority with ↑↓ arrows
• 📊 Progress: Real-time transcription progress with percentage
• 🔄 Refresh: Single unified refresh button updates all data
• 📁 Results: View and download transcripts/SRT stored in database
• 🗒️ Logs: View application logs (last 1000 entries)

Persistent Storage:

• data/uploads/ - Input audio/video files
• data/sqlite.db - SQLite database containing:
  • Job queue (current, queued, completed jobs)
  • Transcription results (transcript text and SRT)
  • Application logs (last 1000 entries)
• data/models/ - Cached Whisper models

⁠Command Reference

⁠Combined Command Table

⁠Environment Variables (Docker)

⁠PowerShell Launcher (run_whisper.ps1)

⁠Docker Launcher (start-docker.ps1 / start-docker.sh)

⁠Python Script (audio_to_text4.py)

⁠Model Selection

GPU Acceleration: 15-30x faster with NVIDIA GPU (auto-detected)

References: OpenAI Whisper Models⁠ | Model Card⁠

⁠File Support

Audio: MP3, WAV, FLAC, AAC, M4A, OGG, AMR
Video: MP4, AVI, MOV, MKV, FLV, WMV
Output: .txt transcripts with timestamps, .srt subtitles (optional)

⁠Support

• GPU Status: Check web UI footer or run nvidia-smi
• Docker Logs: docker logs whisper-webapp
• Issues: Report with error messages and GPU model info