Hervé BREDIN

Too many SPEAKER_XX labels?

4

The screenshot shows the UI before I edit anything. The model identified one speaker (leonard) and left another one unidentified? Why is there so many SPEAKER_XX? (speaker_00 to speaker_06)?

pyannote --citation

3

[`parallel`](https://www.gnu.org/software/parallel/) prints the following message when you run it ``` Academic tradition requires you to cite works you base your article on. If you use programs that use GNU Parallel...

Better handle speakers at chunk boundaries in pyannote.diarization

1

I am wondering whether it would be a good idea to actually overlap chunks a bit. Instead of showing [10, 20] -> [20, 30], we could actually show [9, 21]...

Handle saving/loading of speaker embedding in pyannote.diarization recipe

1

We need to discuss how speaker embeddings can be reused/reloaded when relaunching Prodigy on an existing file. How/where are embeddings stored as numpy array on disk

Logging user interaction in pyannote.diarization recipe

1

To evaluate the impact of various design choices, we should log the user interactions and store them into the Prodigy database. For instance, we should log: * how many times...

Useful multi-label models

7

I'd like to pretrain a couple of models to host then on hugginface along with the others. What kind of classes/train datasets would you suggest? I was thinking about `MALE`/`FEMALE`...

Interactive pyannote.audio.utils.preview

5

Using the following piece of code in a Jupyter Notebook will create a nice video where audio is synchronized with the annotation ```python from pyannote.audio.utils.preview import preview from pyannote.core import...

Investigate torchaudio streaming API

3

https://github.com/pytorch/audio/issues/1442#issuecomment-1032358815

Huggingface Inference API and widget

4

[pyannote.audio 2.0](https://github.com/pyannote/pyannote-audio/tree/develop) will bring a unified pipeline API: ```python from pyannote.audio import Pipeline pipeline = Pipeline.from_pretrained("pyannote/speaker-diarization") output = pipeline("audio.wav") # or pipeline({"waveform": np.ndarray, "sample_rate": int}) ``` where `output` is a...