DESCENT PROTOCOL // BOOTING

Loading Sequence  ·  0%

2026In ProgressSolo backend engineer

audio-transcriber

Self-hosted audio transcription API. faster-whisper for speech, PANNs for sound events, Redis-backed job queue with SSE streaming for progress.

PythonFastAPIfaster-whisperPANNsRedisDockerSSE

Overview

A self-hosted audio transcription service. Uploads are queued in Redis and picked up by a worker that runs faster-whisper for speech plus PANNs for environmental sound event detection. The API streams progress updates back to the client via server-sent events rather than returning a single large blob at the end. Packaged for Docker Compose, CUDA-accelerated when available.

Highlights

  • faster-whisper for speech transcription with CUDA acceleration when available.
  • PANNs sound event detection so output includes non-speech sounds (applause, music, etc.).
  • Redis job queue so long uploads do not block the HTTP layer.
  • Server-sent events stream progress to the client instead of polling.

Challenges and approach

  • Long transcription jobs over HTTP would time out behind proxies. Moving progress to SSE and keeping the final transcript in Redis for retrieval solved both ends of the problem.
  • GPU warmups cost seconds per cold start. Keeping the worker resident and the model loaded between jobs amortized that cost across many requests.