Disband
Overview

Overview

The Disband API splits audio files into stems: upload a file, wait for the separation, download the stems.

How it works

A separation takes three requests: upload the file and name the stems, wait until the separation is ready, then download each stem. The quick start runs all three in one program.

POST /separations Create a separation

ParameterInTypeDescription
stems requiredquerystring[]Stem ids from the catalog, repeated as stems=a&stems=b
formats querystring[]File formats to deliver each stem in, repeated like stems
name querystringThe file's name, for the list and the download file names
bytes queryintegerSize in bytes of a file to upload separately instead of sending it as the request body
body requiredbodyapplication/octet-streamMP3, WAV, FLAC, M4A, Ogg or Opus audio

GET /separations/{id} Get a separation

ParameterInTypeDescription
prefer headerstringwait=N holds the response until the next status change, at most N seconds, up to 60
id requiredpathstringSeparation id

GET /separations/{id}/stems/{stem}/{format} Download a stem

ParameterInTypeDescription
id requiredpathstringSeparation id
stem requiredpathstringStem id from the catalog
format requiredpathstringwav or mp3, one of the separation's formats

Get started

  • Quick start: Split a song into vocals and instrumental and download both stems, in cURL, JavaScript or Python.
  • Authentication: Requests authenticate with an API key in the Authorization header.

Guides

  • Separations: Create a separation, follow its status, download its stems and delete it.
  • Large files: Files over 95 MB are uploaded to a reserved upload URL instead of the request body.
  • Stems: The stem catalog, how stems relate to each other, and silent stems.
  • Credits: How the price of a separation is calculated, and how credits are bought.
  • Errors: The error format, which errors to retry, and every problem type.

Reference

  • Limits: File length and size limits, concurrent separations, and how long results are kept.