Chris Needham

Thanks, Mark. We're certainly interested in pursuing this for V2.

A rough model of different use cases vs accuracy requirements is Figure 1 in [this paper](https://dl.acm.org/citation.cfm?doid=2932206.2932214), shown below: 

The minutes from the joint meeting with Media WG are [here](https://www.w3.org/2022/01/11-mediawg-minutes.html#t02). Summary: * We can use references in the WebCodecs registry for codec string details: https://www.w3.org/TR/webcodecs-codec-registry/

Yes, that could work, but it doesn't really help to know how to implement. Is a fall back enough? If we do need to be more specific, then we'd need...

Thanks! I'll take a look...

For now, I've just added an error message, as it currently fails silently.

This can't really be done at the moment, and would require new options to be added to the library.

The test failures all relate to MP3 decoding, so relate to libmad. For the tests to pass, libmad has to produce the exact same output audio, sample by sample, as...

I'd prefer to figure out how to make the tests consistent. That could mean comparing outputs against platform-specific expected output files or introducing a tolerance.

> How much do the above values increase the pixels or percentages in the graph? > > Can I use the values other than the above and what is the...