Information

Revisiting in-band text tracks in MediaSource Extensions
  • Past
  • Confirmed
  • Breakout Sessions

Meeting

Event details

Date:
– UTC
Status:
Confirmed
Location:
Marigold
Participants:
Alicia Boya Garcia, Wei Ding, Nigel Megitt, Lingjun Meng
Big meeting:
W3C Breakouts Day 2025 (Calendar)

The MediaSource Extensions (MSE) spec has included in-band text tracks since the first published working draft, but that part of the spec have lacked browser adoption for a very long time.

Recently, WebKit has added experimental support for in-band text tracks in MSE. Work on this has highlighted a number of challenges in bytestream formats and the MSE specifications.

The purpose of this breakout session is to gather together, discuss those challenges and come to agreement on the direction to tackle them in the MSE specifications.

Agenda

Chairs:
Alicia Boya Garcia

Description:
The MediaSource Extensions (MSE) spec has included in-band text tracks since the first published working draft, but that part of the spec have lacked browser adoption for a very long time.

Recently, WebKit has added experimental support for in-band text tracks in MSE. Work on this has highlighted a number of challenges in bytestream formats and the MSE specifications.

The purpose of this breakout session is to gather together, discuss those challenges and come to agreement on the direction to tackle them in the MSE specifications.

Goal(s):
Clarify some finer points about how in-band text tracks in MSE should work and define next steps for improvements in the specifications.

Agenda:
Basic familiarity with MediaSource Extensions is expected for attendees.

To make the session easier to follow, the session will start with a very brief introduction of specific formats that will be relevant for the discussions, such as:

  • WebVTT in ISO BMFF (MP4)
  • WebVTT in WebM: S_TEXT/WEBVTT and D_WEBVTT/kind

After that introduction, a number of topics will be discussed. For each topic there is at least one question that we hope to answer.

  • Cues vs samples vs MSE coded frames
    • Should the MSE spec refer explicitly to cues, and if so, how?
  • Would it be desirable for MSE to have a concept of sparse stream gaps? Could the bytestream formats allow it?
    • The SourceBuffer with only a text track problem
    • Use cases other than text tracks
  • Embedded CEA/CTA-608/708 signalling
    • Would support of embedded CTA-708 in MSE be desirable?
    • What would be the best way to accomodate embedded text tracks like this?
  • Cues across segment boundaries: ISO BMFF, WebM and processing in MSE
    • Should the MSE spec mandate cues are extended instead of duplicated?
    • Should the MSE spec mandate that when cues are extended, they are updated and "oncuechange" is emitted, instead of removed and re-added?
  • MSE WebM bytestream: are the current WebVTT representations acceptable for MSE?
    • Should we advocate for any specific improvements?
    • Should the MSE WebM bytestream spec be updated to refer to any specific WebVTT representation?

Interested parties are encouraged to call attention to any other existing challenges with in-band text tracks in MSE and the related technologies.

This list is not final and given the limited time, it may not be possible to cover all the topics, much less agree on answers for most of the questions. These discussions can still serve as a starting point for later work in the working groups.

Slides: https://ntrrgc.github.io/w3c-breakouts-2025-mse-text-tracks/

Materials:

Minutes

 Read minutes

Export options

Personal Links

Please log in to export this event with all the information you have access to.

Public Links

The following links do not contain any sensitive information and can be shared publicly.

Feedback

Report feedback and issues on GitHub.