juanmc2005 / diart

A python package to build AI-powered real-time audio applications
https://diart.readthedocs.io
MIT License
1.06k stars 87 forks source link

Create a documentation page #75

Closed juanmc2005 closed 11 months ago

juanmc2005 commented 2 years ago

Problem

As the codebase grows, it is difficult for a user to keep track of all the features and mix-and-match possibilities available.

Idea

Create a documentation page with tutorials and an API reference.

amritbhat786 commented 2 years ago

Yess please. There are parts in the current documentation page which seem slightly disconnected. For eg, in the [Add your model] (https://github.com/juanmc2005/StreamingSpeakerDiarization#add-your-model) section, as a first timer I do not understand how to run the code. Specifically because the code loads a "my_model.ckpt" file but I don't know how to create this file. It would be better if this model file could either be referenced in the repo under an "examples" folder or if its very huge, be downloadable for the demo code to run. To summarize, the example code should be runnable off the shelf (preferably without any modifications to the code itself.)

juanmc2005 commented 2 years ago

Hi @amritbhat786,

The "Add your model" section of the README is for custom models (think EEND or ECAPA-TDNN for example). The default pipeline uses pyannote/segmentation and pyannote/embedding automatically and you can run it with diart.stream or with RealTimeInference like it's shown in the README.

On the other hand I think the "Add your model" title is indeed a bit misleading, do you think it would be clearer if it was something like "Custom models"?

amritbhat786 commented 2 years ago

Hi, thanks for the clarification. Yes, custom models sounds good to me. Also a comment stating that a file path should be specified in the example script would help as well.

juanmc2005 commented 2 years ago

Thanks @amritbhat786 ! I'll make sure to make those changes for the next release. If you see any other clarifications that you'd like to make in the readme, you can directly create a pull request. That way it's easier for me to quickly accept and include them in the next release :)

juanmc2005 commented 11 months ago

A first version of a documentation page is already in develop and will appear on readthedocs with the v0.9 release. I'll create a separate issue for tutorials