Mexca

Fixes mismatch between PyPI release version and GitHub release version.

Fixes one crucial and one minor bug.

Changed

- Output files in the standard pipeline recipe are save after each video file is processed (instead of saving everything at the end)
- GitHub action workflows are tested on MacOS version 13 because FFMPeg cannot be automatically installed on newest version
- An extra step for freeing disk space is added to GitHub action Docker workflows

Fixes

- A bug in the `FaceExtractor` component where the input for the `MEFARG` model was not sent to the correct device (GPU only)
- A bug in the `SentimentExtractor` compoenent where the tokenizer would raise a run time error in very rare cases (probably for very long sentences). Now, padding is added to avoid the error and exceptions are caught returning `null` sentiment scores.

Changes the pretrained MEFARG model to be downloaded from Hugging Face Hub instead of Google Drive due to problems with the gdown package.

Fixes the PyPI publication. PyPI could not publish the previous version due to a name duplication error.

Contains some final fixes and adjustments for the first complete release.

Changed

- Upgrades pyannote.audio to version 3.1.1
- Downgrades gdown to version 4.6.0
- Only essential steps are logged on INFO level (i.e., cluster confidence, average embeddings, and removing audio files is now on DEBUG level)
- The error message when the connection to the Docker daemon fails is now more informative

Removed

- onnx-runtime, ruamel.yaml, and torchaudio as requirements for the speaker identifier component due to pyannote.audio upgrade

Fixed

- A bug caused by pyannote.audio version 3.0.0 for short audio clips when frame-wise detected speakers exceeded maximum number of speakers (see 106)
- An issue by gdown when model files hosted on Google Drive could not be accessed anymore (https://github.com/wkentaro/gdown/issues/43)

Adds average speaker embeddings and improved speaker diarization. Also increases the performance of data processing. Provides an advanced example notebook for extending the standard MEXCA pipeline.

Added

- The `SpeakerAnnoation` class has a new attribute `speaker_average_embeddings` containing the average embeddings for each detected speaker
- The `SpeakerIdentifier` has a new argument to explicitly set the device its run on (by default CPU)
- The `SpeakerIdentifier.apply()` method has a new `show_progress` argument to enable progress bars for detected speech segments and embeddings
- A new notebook on customizing and extending the MEXCA pipeline (`examples/example_custom_pipeline_components.ipynb`)
- Two new recipes for applying the standard MEXCA pipeline and postprocessing the extracted features (`recipes/`)
- The `Pipeline.apply()` method has a new `merge` argument to disable merging features from different modalities; this is useful when customizing a pipeline
- A new logo (thanks to [Ji Qi](https://github.com/jiqicn))
- Documentation on how to use mexca with GPU and CUDA support
- notebook has been added as a dependency for the demo installation
- scikit-learn has been added as an explicit dependency (previously dependency of py-feat)

Changed

- pyannote.audio has been upgraded to version 3.0.0; this required adding the following dependencies:
- torch >= 2.0.0
- onnxruntime-gpu on Windows and Linux
- onnxruntime on MacOS
- torchaudio on MacOS
- torch has been upgraded to version 2.0.0 for all components requiring it
- The `SpeakerIdentifier` component uses the `pyannote/speaker-diarization-3.0` model by default
- pandas has been replaced by polars; the `Multimodal.features` attribute now stores a `polars.LazyFrame` instead of a `pandas.DataFrame`; this speeds up postprocessing and merging for large data sets

Removed

- py-feat has been removed as a dependency