GitLab

* feat: efficient backprop for a_coeffs

* update lfilter docstring

* doc: double quotation

* use dafx paper for reference as it includes more details

* revise docstring

Summary:
Pull Request resolved: https://github.com/pytorch/audio/pull/3803

The model checkpoint path can not be created for Squim models. Use the latest download_asset method to fix it.

Reviewed By: moto-meta

Differential Revision: D59061348

Differential Revision: D57294298

Pull Request resolved: https://github.com/pytorch/audio/pull/3791

[codemod] c10::optional -> std::optional in pyspeech/experimental/csrc/decoders/TransducerDecoder.h +20

Differential Revision: D57294284

Pull Request resolved: https://github.com/pytorch/audio/pull/3793

[codemod] c10::optional -> std::optional in pytorch/audio/src/libtorio/ffmpeg/stream_reader/stream_processor.h +20

Differential Revision: D57294285

Pull Request resolved: https://github.com/pytorch/audio/pull/3792

The lengths of targets and log_probs should be reversed.

* add golf and dynonet paper

* doc: add references

* add EOF

* fix: line too long

* remove line end space

* remove indentation
Co-authored-by: moto <855818+mthrok@users.noreply.github.com>

---------
Co-authored-by: moto <855818+mthrok@users.noreply.github.com>

Addresses #3688 

When the input is zero Tensor, the result should be empty.

Differential Revision: D50696105

Pull Request resolved: https://github.com/pytorch/audio/pull/3682

Differential Revision: D50677606

Pull Request resolved: https://github.com/pytorch/audio/pull/3681

Differential Revision: D50633306

Pull Request resolved: https://github.com/pytorch/audio/pull/3675

Add torio top-level directory. It's not part of the package yet.

Differential Revision: D50602496

Pull Request resolved: https://github.com/pytorch/audio/pull/3670

Differential Revision: D50506299

Pull Request resolved: https://github.com/pytorch/audio/pull/3669

So that release engineering team does not have to manually remove prototype-related things
in each release

Differential Revision: D50367721

Pull Request resolved: https://github.com/pytorch/audio/pull/3662

The `encoder_layer_norm_first` should be set to True for the Large model of WavLM.
Address https://github.com/pytorch/audio/issues/3347

Summary:
Revamped version of https://github.com/pytorch/audio/pull/3234
(which was also revamp of https://github.com/pytorch/audio/pull/2850)

Differential Revision: D50197331

Pull Request resolved: https://github.com/pytorch/audio/pull/3654

Differential Revision: D50193749

Pull Request resolved: https://github.com/pytorch/audio/pull/3650

Differential Revision: D50205775

Pull Request resolved: https://github.com/pytorch/audio/pull/3651

Differential Revision: D50229857

Pull Request resolved: https://github.com/pytorch/audio/pull/3652

Differential Revision: D50086556

Pull Request resolved: https://github.com/pytorch/audio/pull/3648

Differential Revision: D50082877

Pull Request resolved: https://github.com/pytorch/audio/pull/3646

Addresses https://github.com/pytorch/audio/issues/3640

Differential Revision: D49965263

Pull Request resolved: https://github.com/pytorch/audio/pull/3639