I re-generated these using opus_encode instead of opus_encode_float which reduces the number of changed .iamf bitstreams. But since we released v1.0.0 bitstreams we have found or fixed other subtle bugs in the encoder.

For example one of these bugs was applying the incorrect demixing parameter in some circumstances. The audio frames would have been slightly off the correct results. This was not caught in conformance testing because we were doing some kind of SNR-based test and overall the audio would have been close.

I think it is better to take these changes anyway.

Choose one of this PR or #80 and then close the other without merging it in.

Accepting either of the related PRs are fine. If you already performed validation on the other one we can take it instead.

Accepting either of the related PRs are fine. If you already performed validation on the other one we can take it instead.

We will perform validation on this PR