#531: WebRTC Insertable Streams

Visit on Github.

Opened Jun 29, 2020

Hello TAG!

I'm requesting a TAG review of WebRTC Insertable Streams.

WebRTC Insertable Streams is a proposed API that allows the insertion of user-defined processing steps in the pipeline that handles media in a WebRTC context. It defines a number of extensions to objects defined in WEBRTC-PC, and it draws upon some definitions originally proposed for WEB-CODECS.

Further details:

  • I have reviewed the TAG's API Design Principles
  • The group where the work on this specification is currently being done: W3C WebRTC Working Group
  • The group where standardization of this work is intended to be done (if current group is a community group or other incubation venue): W3C WebRTC Working Group
  • Major unresolved issues with or opposition to this specification: None for the API itself, but more work needed for future modules (e.g., end-to-end encryption module, support for raw media frames) to better support use cases.
  • This work is being funded by: Google

We'd prefer the TAG provide feedback as (please delete all but the desired option): 💬 leave review feedback as a comment in this issue and @-notify @guidou and @alvestrand

Discussions

2020-07-20

Minutes

Dan: [reviewing explainer]

Yves: they are using transform streams.. seems good.

Sangwhan: the pattern stuff is fine, the use cases are fine... does it interact with anything other than WebRTC? The answer would seem to be no.

Yves: possibility to hide computation-intensive processing here... bitcoin mining.

Sangwhan: why do we not have video/media worklets? Audio has audio worklets. WebRTC has their own thing. If you want to do video preprocessing in a worker or service worker there's no nice way to do it except at bitstream level...

Dan: Where could it be?

Sangwhan: HTML media...

Yves: transform streams...

Sangwhan: latency requirement was a real problem. you don't get the latency guarantee.

[bumped

2020-07-27

Minutes

[punt to plenary

2020-08-17

Minutes

Yves: We were waiting for Sangwhan's feedback, Tess added herself to the list of reviewers.

Sangwhan: I remember looking at this briefly... seemed ok to me.

Yves: Some question about why audio worklet over streams... mostly historical... not sure if they would choose streams now. Might make sense to use streams now instead?

Sangwhan: I don't have strong objections. Unless Tess or Yves have strong objections..

Yves: I'm ok to propose close and ask Tess at the plenary if she agrees.