Coalesce

Coalesce is an audio editor which makes slicing dialogue as easy as editing text.

Project Status

🚧 Alpha 🚧

Features:

• AI transcription using faster-whisper
• Nondestructive text editing: remove and reorder spoken words as text
• Visually refine word timings with a waveform editor
• Export mixed down mono 48khz audio

Next up:

• [x] Lazy load audio data from chunks
• [x] Improve playback perf w/ incremental lookahead audio scheduler
• [x] Highlight words as they're played
• [x] Drag and drop files to transcribe and process
• [x] Collaborative editing
• [x] Export separate audio tracks
• [ ] Add sound clips

How to use

The easiest way to get started is to build and launch the containers:

1. DOCKER_BUILDKIT=1 docker-compose --env-file docker-compose.env --profile process-audio --profile mailslurper up
2. Browse to https://localhost:3333
3. Access registration emails at https://localhost:4436

Development

To run all services in watch mode:

DOCKER_BUILDKIT=1 docker-compose --env-file docker-compose.env -f docker-compose.yml -f docker-compose.dev.yml --profile process-audio --profile mailslurper up

Tailing logs:

docker-compose -f docker-compose.yml -f docker-compose.dev.yml logs --follow