minor fix (#1418)

Fix to fixed batch size bucketing and audio loading network connectio… (

#1387)

* Fix to fixed batch size bucketing and audio loading network connection resets

* Fix tests and add more 'paranoia' tests

Refactor bucket selection for customization (#1377)

* Refactor bucket selection to allow customization

* Extend the API further

* Prune imports

augmentation/torchaudio: add Phone effect (mulaw, lpc10 codecs) (#1348)

* augmentation/torchaudio: add Phone effect (mulaw, lpc10 codecs)

* restore_orig_sr option

---------

Co-authored-by: Piotr Żelasko <[email protected]>

Releease 1.24.2

Support for reading data from AIStore using Python SDK (#1354)

* Support for reading data from AIStore using Python SDK

* More AIStore related docs

Add new sampler: weighted sampler (#1344)

* add file

* add a weighted data source to enable sampling based on per-sample weight; do not allow duplicated sample within the same epoch

* add a weighted sampler; do not allow lazy mode; do not allow duplicated cut in the same batch

* modify init file accordingly

* add more documentations

* use numpy for sampling; pre-compute the indexes in __iter__ to save time

* add more documentation

* minor changes to the arguments

* remove unused file

* add test

* add more docs

* fix isort

* inherit from SimpleCutSampler; remove duplicated code

* minor fix

* Add changes requested in code review

---------

Co-authored-by: Piotr Żelasko <[email protected]>

In CommonVoice corpus, use .tsv headers to parse and not column index (…

…#1328)

* Fix for cv corpus

* Fix for cv corpus x2

* Debug serialization problem

* Debug serialization problem

* Undo

* Handle quote polution in CV dataset

Bump dev version to 1.23.0 (#1301)

`AudioBackend` specific `save_audio` and `info`, managing missing SoX…

… in torchaudio, Python 3.12 / PyTorch 2.2 support, using `libsndfile` as preferred audio backend (#1288)

* AudioBackend supports save_audio() [code cleanup]

* Fix for Path handling; skip some tests on older torchaudio

* Move info() implementation to each AudioBackend

* Update CI configurations, fix more tests

* Conditionally build kaldifeat, fix some more tests

* Fix more tests, remove dead code, bump version

* Remaining fixes, legacy OPUS reading mode env var

* Fixes for torchaudio==2.0.0

* Fixes for save_audio/save_audios

* Skip backends for audio saving that are not applicable (e.g. torchaudio backends when it's not installed)

* Prefer LibsndfileBackend as the default lhotse backend (except for some special cases) + fix CutSet.copy_data()