search

ina-foss / inaSpeechSegmenter

CNN-based audio segmentation toolkit. Allows to detect speech, music, noise and speaker gender. Has been designed for large scale gender equality studies based on speech time per gender.
MIT License
717 stars 127 forks source link

Provide draft code for training a model #62

Closed realies closed 2 years ago

realies commented 2 years ago

There are a lot of issues related to questions on how to train a custom model, and requests for a script to train own models.

Would it be possible for the R&D team behind this project to provide the draft code used to train its model in some form, without the actual dataset? It would be very useful for creating a contribution adding training code for inaSpeechSegmenter.

https://github.com/ina-foss/inaSpeechSegmenter/issues/48 https://github.com/ina-foss/inaSpeechSegmenter/issues/41 https://github.com/ina-foss/inaSpeechSegmenter/issues/12 https://github.com/ina-foss/inaSpeechSegmenter/issues/5 https://github.com/ina-foss/inaSpeechSegmenter/issues/4

DavidDoukhan commented 2 years ago

The amount of word required to provide the training code is too important and I'm lazy This may be done for the release of future inference models. Kind regards,

realies commented 2 years ago

@DavidDoukhan, that's great... :):) Would you mind sanity-checking an implementation of the training code if there's a PR?