Stable-ts

* `WhisperHF.transcribe()` can now take generation parameters supported by `Transformers` ([133f323](https://github.com/jianfch/stable-ts/commit/133f3235c34aced23c95b10a37df3031b6e4cf0f))
* added logic to replace `None` timestamps returned by Hugging Face Whisper models ([8bbe0c5](https://github.com/jianfch/stable-ts/commit/8bbe0c59bf3125df310596ec0dfecef783b3b437))
* changed `whisper_word_level.hf_whisper.load_hf_pipe()` model loading method([a684fb4](https://github.com/jianfch/stable-ts/commit/a684fb4f5b271079b243b32a9552426340419806))

* added DeepFilterNet (https://github.com/Rikorose/DeepFilterNet) as supported denoiser ([3fafd04](https://github.com/jianfch/stable-ts/commit/3fafd04902666675bb0e5786c8a08ee18b68f908))
* added Whisper on Hugging Face Transformers to CLI ([3fafd04](https://github.com/jianfch/stable-ts/commit/3fafd04902666675bb0e5786c8a08ee18b68f908))
* fixed CLI throwing OSError when input is a URL and --output is not specified ([3fafd04](https://github.com/jianfch/stable-ts/commit/3fafd04902666675bb0e5786c8a08ee18b68f908))
* fixed `WhisperHF.transcribe()` unable to load when audio is URL or certain formats ([3fafd04](https://github.com/jianfch/stable-ts/commit/3fafd04902666675bb0e5786c8a08ee18b68f908))

* added support for Whisper on Hugging Face Transformers ([9197b5c](https://github.com/jianfch/stable-ts/commit/9197b5c92b31d67b58e2c1bbf8e5283cd7faf4cb))
* fixed non-speech suppression not working properly for `transcribe_any()` ([9197b5c](https://github.com/jianfch/stable-ts/commit/9197b5c92b31d67b58e2c1bbf8e5283cd7faf4cb))

* changed default to `dtype=numpy.int32` for all Numpy int arrays ([3886bc6](https://github.com/jianfch/stable-ts/commit/3886bc6932a91784d8229deb3e811f75d8ea122e))

* removed `shell=True` in `.audio.utils.get_metadata()` ([e8f72a3](https://github.com/jianfch/stable-ts/commit/e8f72a3b27ddb14202d50f9eaf4af7afccadbaa2))

* added "「" to `prepend_punctuations` and "」" to `append_punctuations` ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* added `AudioLoader` class for handling general audio loading ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* added `NonSpeechPredictor` class for handling non-speech detection ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* added `default.py` to hold global default states ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* added `failure_threshold` to `align()` ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* added `stream` to functions that use `AudioLoader` internally ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* added progress bars for VAD and Demucs operations ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* changed text normalization for `align()` ([6d0746c](https://github.com/jianfch/stable-ts/commit/6d0746cd79ece5aea0e08488da1687ef4fe65935))
* changed `WhisperResult` to ignore segments with no words ([6d0746c](https://github.com/jianfch/stable-ts/commit/6d0746cd79ece5aea0e08488da1687ef4fe65935))
* changed `nonspeech_error` default from 0.3 to 0.1 for all functions ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* changed `nonspeech_skip` default from 3.0 to 5.0 for `align()` ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* changed `use_word_position` behavior ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* changed to load Demucs into cache for reuse by default ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* deprecated and replaced `demucs` and `demucs_options` with `denoiser` and `denoiser_options` ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* dropped `ffmpeg-python` dependency ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* dropped dependencies: more-itertools, transformers ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* fixed `align()` producing empty word slices ([6d0746c](https://github.com/jianfch/stable-ts/commit/6d0746cd79ece5aea0e08488da1687ef4fe65935))
* fixed `refine()` exceeding the max token count ([297](https://github.com/jianfch/stable-ts/issues/297)) ([f6d61c2](https://github.com/jianfch/stable-ts/commit/f6d61c228d5a00f89637422537d36cd358e5b90d))
* fixed issues in `transcribe_any()` caused by unspecified samplerate ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* fixed `vad=True` causing first word of segment to be grouped with previous segment ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* refactored `audio.py`, `stabilization.py`, `whisper_word_level.py` into subpackages ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))
* removed `demucs_output` ([9968a45](https://github.com/jianfch/stable-ts/commit/9968a4510b2973be40e75d7978c915f3727ed06f))