Speaker 1: Hi there, apologies for me sounding like I'm so sick. I am so very sick. I have a flu, got it in a flight. But in this video, I want to share with you something that I found related to OpenAI Whisper. Someone went and converted the models to basically machine learning models that run really well on MacOS, on Apple Silicon, and on Intel Macs. It runs really fast. I'm able to transcribe a 30-minute audio in about two minutes and have all the files. So I'm going to show you now how you can get that on your computer and use it. So here we go. First of all, this is the actual project is from G. Gerganov. And all of my shortcuts are going to be pulling from here, okay? So I'm going to leave this link here for you on the descriptions. There will be on the descriptions a link to the shortcuts as well. So you can install on your computer. First thing, after you downloaded the shortcuts, you're going to have two shortcuts. You're going to have this Install Whisper AI M1 version 2, and you're going to have the Transcribe English Tiny version 2, okay? First thing that you want is to execute Install Whisper AI M1 version 2. And just to follow along what's happening, we're going to open Finder on your user folder and execute that shortcut. So what the shortcut will do is we'll create this folder, Whisper. And then inside it, Whisper.cpp. And here it's going to start downloading all of the dependencies that are required. The biggest amount of downloads will be here for the models. So you're going to see them coming along here. We already have now the Base English, we have Tiny, we have Tiny English, Medium is currently being downloaded, and we can see here the speed in which it's downloading. So this is going to take about a couple of minutes, maybe up to five minutes. It really depends on your connection. What you will want to wait for before you can transcribe something is for a main file to appear in this folder. So while this is being executed, the other thing that you're going to need is a compressor profile or use the transcoder that you prefer. Your WAV files, you're going to need to feed WAV files to Whisper, this version of Whisper anyway. And those WAV files will need to be in 16 kilohertz and they will need to be 16-bit. So here it is, sample rate 16, the sample size 16 bits. I will also give you a link to download this compressor profile in case you're not able to create your own. Now we can see here that the installation finished as we have this file main here. Now all we need to do is to execute the transcribe English Tiny version 2. Okay, so here I have already my 16 kilohertz file. And I will just point it to my shortcut and then execute. And this is about four minutes, five minutes video, I think. Let me see. Yeah, it's three and a half minutes. And we're going to see that very quickly. We're going to have our transcription here. And I'm leaving this happen in real time just for you to have an idea how fast it is. Okay, just finished. I have my text file VTT SRT and this extra document file that I put there to be created. And I have my whole transcription, my whole subtitles here. Well, I hope this has been useful for you. And of course, it does help a lot if you subscribe and like this video and share with others. But I think these things are only going to continue getting better and better. So enjoy it.