Show HN: Read any YouTube video with punctuations added using AI in the browser (appblit.com)

7 points by ldenoue 2y ago ↗ HN

12 comments

[ 4.7 ms ] story [ 40.4 ms ] thread

Most YouTube transcripts lack punctuations. So this app adds them back using a deep learning model based on DistilBert converted to ONNX for inference in your browser.

skeptrune 2y ago ↗

I saw it was running within the browser. That's beyond awesome.

ldenoue 2y ago ↗

Yes I’m also surprised how many tasks can be done using ONNX in the browser. The model I found and converted for browser use is here https://huggingface.co/ldenoue/distilbert-base-re-punctuate

skeptrune 2y ago ↗

Wow this is one of the most useful foundational AI things I have seen in a while. Is this open source? I have a really core use-case I think.

ldenoue 2y ago ↗

It’s not open source but I did publish the ONNX converted model to HuggingFace here https://huggingface.co/ldenoue/distilbert-base-re-punctuate

It’s amazing what Xenova Transformers JS can do to run ONNX models. And WebGPU is coming soon!

samspenc 2y ago ↗

Very cool, curious if you plan to add a summarize feature along with this? I know there are other apps that do that already, but would be good to follow along inside the same single app.

ldenoue 2y ago ↗

I would like first to generate chapters (perhaps using a T5 model). Summarization is also possible.

carlosjobim 2y ago ↗

I tried it and it worked great, congratulations! Looking forward to having chapters or at least separate paragraphs, I think you're doing right by giving that priority, as you mentioned in another comment.

ldenoue 2y ago ↗

Chapters should come first then?

carlosjobim 2y ago ↗

Is it possible to get the chapters directly from YouTube? As many videos already have them defined by the creator.

boopbeepbop 2y ago ↗

This is great! Do you rely on YouTube captions or do some sort of ASR?

ldenoue 2y ago ↗

It uses the existing YouTube transcript and adds punctuations using a deep net that runs in your browser