[v2.0.1] Release Tracker

[nateanl]

This issue is for tracking TODO lists which are supposed to be cherry-picked to the release branch. Following is release branch for the 2.0.1 release.

[mthrok]

https://github.com/pytorch/audio/pull/3230 resolves https://github.com/pytorch/audio/issues/3218, which existed since earlier release

---

• [x] https://github.com/pytorch/audio/pull/3238

[mthrok]

https://github.com/pytorch/audio/pull/3204 fixes issue with multiple formats, which existed since the initial release of StreamReader.

---

• [x] #3239

[mthrok]

https://github.com/pytorch/audio/pull/3243 and https://github.com/pytorch/audio/pull/3248 fixes issue with FFmpeg backend which was introduced in 2.0

---

• [x] https://github.com/pytorch/audio/pull/3245

[nateanl]

3252 and #3265 fix WavLM model when it is called in torch.no_grad().

---

• [x] https://github.com/pytorch/audio/pull/3264

[nateanl]

3253 improves Wav2Vec2/HuBERT by using scaled_dot_product_attention to increase efficiency.

---

• [x] https://github.com/pytorch/audio/pull/3261