Aurora Cramer

Isolated source audio files occasionally are longer in duration than soundscape duration

13

Occasionally, when generating mixtures using only foreground events (no background) and saving the isolated events to disk, one of audio files for the isolated sources is slightly longer than the...

https://docs.github.com/en/actions/learn-github-actions/migrating-from-travis-ci-to-github-actions

So we can have a more flexible regression check for necessary changes that change the model output. May be useful (thanks @jonnor!): https://cml.dev/

The linear frequency spectrogram frontend results in 197 frames instead of 199, which seems to be caused by: - In `_librosa_linear_frontend` (`openl3/core.py`), `center=False` should be `center=True` when calling `librosa.stft` -...

Make an object-oriented interface for embedding models

4

An object interface for extracting embeddings would make it simpler to extract multiple embeddings and would allow for the embedding model to just be loaded once instead of every time...

Supporting multiple GPU models

4

Should supporting running the embedding models on multiple-GPUs be prioritized? Here are the pros/cons as I see it (not necessarily equally weighted in terms of importance): ## Pros - Allows...

@iranroman Since STARSS22 was added to Soundata, the DCASE 2022 Challenges ended and the evaluation datasets have been released, and can now be added to the dataset REMOTES and index...

The [stable docs](https://soundata.readthedocs.io/en/stable/source/quick_reference.html) appear to have all of the datasets in the repo, but the [latest ](https://soundata.readthedocs.io/en/latest/source/quick_reference.html) seem to be missing some of those datasets for some reason.

Noticed a small typo I've found in the definitions of `TIME_UNITS`: "miliseconds" should be "milliseconds"