Skip to content
VoxNotch

Self-hosted transcription

Run transcription where the recordings already live.

Self-hosted transcription is not just a local model. It is the full operating path: ingest, queue, transcribe, review, export, back up, restore, monitor, and upgrade without making sensitive audio a vendor dependency.

Stack
Docker, Postgres, Redis, API, worker
Default STT
Local faster-whisper
Storage
Local volume, MinIO, S3, or R2
Recovery
Backup, verify, restore

Self-hosted transcription stack with worker, database, storage, and transcript output.

How to evaluate

Prove the stack before production audio.

  1. Run local demo

    Start the local stack with generated secrets and dynamic ports.

  2. Transcribe one file

    Upload a real recording and inspect speakers, timing, search, and exports.

  3. Verify backup

    Run the backup verification path against scratch state.