From the course: Rust LLMOps

Unlock this course with a free trial

Join today to access over 23,100 courses taught by industry experts.

Whisper Candle transcriber

Whisper Candle transcriber - Rust Tutorial

From the course: Rust LLMOps

Start my 1-month free trial Buy for my team

Whisper Candle transcriber

“

- [Instructor] One of the emerging new things that people are looking at is using pre-trained models for automatic speech recognition. So whisper-large is a good example of something you could use without any fine tuning. And you could actually go in here and look at the different kinds of parameters. For example, maybe a medium would be 769 million parameters or a large, whatever it is you're trying to do. Now, the question though, when you're using tools like this is how do I run it? Well, fortunately with Hugging Face Candle here, you can see here that this framework is a minimalistic high-performance framework that includes GPU support. And in fact there's examples of using Whisper right here. So in order to do that, let's dive into the source code a little bit, look at Whisper, look at the main. Notice that it's actually using clap which is a command line tool parser, and we can look at some of the parameters that are both constants or the ones that can be passed in here as well.…

Contents