Whisper Paper Explained: Robust Speech Recognition via Large-Scale Weak Supervision

❤️ Support the channel ❤️

Paid Courses I recommend for learning (affiliate links, no extra cost for you):

✨ Free Resources that are great:

💻 My Deep Learning Setup and Recording Setup:

GitHub Repository:

✅ One-Time Donations:

▶️ You Can Connect with me on:

Timestamps:
0:00 - Introduction
1:27 - Abstract
2:30 - Introduction
8:02 - Dataset collection and processing
10:29 - Model approach
15:25 - Figure of model
20:25 - Experiments and Evaluation
26:17 - Long form transcription, messy :/
28:36 - Model and Dataset scaling
31:25 - Long form transcription (cont), messy :/
32:32 - Ending

[vishalgoklani]

A fine-tuning example would be awesome! btw, I extracted the audio from this YouTube video and ran it through Whisper, it's not bad. Unfortunately I got back one giant blob of text. We need to use another LLM/transformer to rewrite the output into proper paragraphs, and also to summarize and remove extraneous content: "What is going on guys? Welcome back to another video. In this one, we're taking a " :)

[aniketsingh]

I have few question and ways to use whisper is there any community which talk about it

[champagnebulge]

I can reach the point where Whisper transcribes a YouTube vid, but I'm having trouble with the speaker identification or diarisation part. Anyone got a link to a great tutorial?

[LoneRanger.]

New to your channel. I like that you go into the detail of the papers. Really looking forward to the fine tuning video next. Subscribed. 😊

[gtgs]

Can you do a code and paper walk thru for continuous or binary hopfield pattern storage and recall

[daniellewis]

You are one of the OGs. Thanks Aladdin.

[normalhuman]

Watching your Unet video. This one will be next in line. So much to learn in AI. I keep getting confused. How do you make a mind map of everything and organize all the info? Should explain in a video.