audioproxy

Audioproxy renders audio variants on demand instead of pre-generating them. A signed URL describes what you want (format, bitrate, trim, fades, loudness normalisation, waveform peaks) and the proxy renders it when it is first requested, streams it while it encodes, and caches it for range-capable serving. No job queue, no variants table, no backfill. One container, no database, reads from a mounted directory or any S3-compatible bucket. If you know imgproxy, this is that, for audio.