Hervé BREDIN comments

Results 270 comments of


                                            Hervé BREDIN

Cannot apply data augmentation

Would you mind contributing a PR that updates the `training_a_model.ipynb` notebook accordingly?

Segmentation fault on import Pipeline when installed alongside transformers >= 4.0 from pip

Would you mind sharing a link to a Google Colab containing a minimum reproducible example that I can quickly (understand: in less than 5 minutes) run to reproduce the error?

Issue with segmentation model

You may try to load the previous version using `pyannote/segmentation@Interspeech2021` instead of `pyannote/segmentation`.

Issue with segmentation model

Would you mind sharing a link to a Google Colab containing a minimum reproducible example that I can quickly run to reproduce the error?

I can think of several others: * `SPEECH` vs. `MUSIC` vs. `NOISE` with [`MUSAN`](https://www.openslr.org/17/) or [`AVASpeech - SMAD`](https://github.com/biboamy/AVASpeech_Music_Labels) or [`TVSM`](https://github.com/biboamy/TVSM-dataset) * Stuttering event detection with [Sep-28k](https://machinelearning.apple.com/research/stuttering-event-detection) * `Sheldon` vs. `Leonard`...

Useful multi-label models

* sound event detection with [AudioSet](https://research.google.com/audioset/) (and its [strong labels](https://research.google.com/audioset/download_strong.html)) * [VGGSound](https://www.robots.ox.ac.uk/~vgg/data/vggsound/)

Useful multi-label models

Laughter detection

Useful multi-label models

[PodcastFillers](https://podcastfillers.github.io/)

Useful multi-label models

[LibriStutter](https://borealisdata.ca/dataset.xhtml?persistentId=doi:10.5683/SP3/NKVOGQ&version=1.0)

Ctrl ne fonctionne pas

I am afraid that we will have to go this way, indeed. Maybe instead of CTRL+`key`, we could go with `key` directly. The one problem I see is finding two...