Difference between revisions of "Proposal Source Separation"

From Audacity Wiki
Jump to: navigation, search
(Interface: added link text)
(Not GSoC (at least not this version anyway))
Line 42: Line 42:
[[Category:GSoC]] [[Category:Feature Planning]]
[[Category:Feature Planning]]

Latest revision as of 20:24, 19 February 2021

Proposal pages help us get from feature requests into actual plans. This page is a proposal for a feature that will allow parts of a single audio stream to be separated. The focus is on the UI for this.
Proposal pages are used on an ongoing basis by the Audacity development team and are open to edits from visitors to the wiki. They are a good way to get community feedback on a proposal.

  • Note: Proposals for Google Summer of Code projects are significantly different in structure, are submitted via Google's web app and may or may not have a corresponding proposal page.

Project Proposal

  • Source Separation is the opposite of mixing. Given a track with multiple sources on it, how do you get to the separate sources? There's no general algorithm to do it, but some cases are not impossibly hard, such as separating a flute from background (audience) noises.
  • Source Separation and audio classification are closely related and the proposal is to tackle the interface for both at the same time. A classifier can be seen as a source separator where of the tracks separated all but one are silent at each sample.

Developer Backing

Use Cases

  • Silence finders: For use in truncate silence effect.
  • Radio phone-in interview: Separate and label the two speakers based on the frequencies. Telephone has much reduced bandwidth providing a simple criterion on which to separate.

Any one of these would be a significant advance for Audacity.


Is Source Separation Hard?

Yes :-) Some of the difficulties are....

On the plus side....


  • The commercial program... (possibly compare to software for blue screening)
  • See ISSE (an interactive source separation editor) for ideas.