audioproxy

What it is

audioproxy: the homepage

audioproxy is an open-source, self-hosted audio transcoding proxy similar to imgproxy but for audio files. It accepts signed URLs that describe audio variants, including format, bitrate, trimming, loudness normalization, and waveform peaks, then renders these variants on demand and caches them for range-capable serving.

The proxy supports multiple output formats including MP3, Opus, Ogg, AAC, M4A, FLAC, and WAV. It can trim audio, apply fades and gain, normalize loudness to EBU R128 standards, and generate audiowaveform-compatible waveform peaks. The URL encodes all options as key-value segments that fully describe the output and serve as the cache key.

Built in Elixir and available as a Docker container, audioproxy streams rendered variants to the first requester while writing to cache for subsequent range-capable access. It enforces audio-only content and is designed to be integrated with storage backends like S3 or local directories.

Quick answers

1. What formats does audioproxy support?

   audioproxy supports MP3, Opus, Ogg, AAC, M4A, FLAC, WAV output formats, and can also generate waveform peaks in JSON or binary format.

2. How does caching work in audioproxy?

   The first request streams the rendered variant to the client while simultaneously writing it to cache. Subsequent requests for the same cache key are served as range-capable 302 redirects to presigned variant URLs, enabling seeking and efficient CDN serving.

3. What is included in the URL options?

   Options in the URL describe format, bitrate, trimming points, fades, gain, loudness normalization, waveform peaks, expiration time, and other processing parameters. All options are encoded as key-value segments that fully describe the output.

4. Can audioproxy handle video files?

   No, audioproxy enforces audio-only content and refuses any source carrying a genuine video stream with a 415 response before rendering begins.

5. What audio processing is available?

   audioproxy offers trimming, fades, gain adjustments, EBU R128 loudness normalization, and a speech cleanup preset that applies high-pass filtering, denoising, de-essing, compression and limiting.