Speaker 1: What is going on, guys? Welcome back. In this quick video today, we're going to learn how to easily transcribe audio files using OpenAI Whisper in Python. So let us get right into it. All right, so this video is going to be very short because the process is extremely simple and we only need a few lines of code to make this work. And essentially, the idea is that we have an audio file. For example, here I have this videosound.mp3, which is basically just the audio file or the audio data for one of my videos. We can play this here real quick. It's essentially just one of my videos as an audio file. And what we want to do now is we want to transcribe this professionally. We want to get the text, for example, to create subtitles or to just do some machine learning work to create a search engine for my videos. Maybe I want to transcribe all of my videos and I want to be able to create a search engine so that you can look for a term and then you can find where it occurs in a certain video. That would be one use case. But we want to do that professionally and we don't want to do this with a lot of mistakes. Now, it's not going to be perfect because there are certain words that are just not going to be recognized. So, for example, certain package names, certain names of applications and technologies that are not necessarily in the dictionary, those are going to be a little bit more difficult to transcribe. But essentially, this is going to be a very high quality transcription. So, what we're going to do is we're going to create a new Python file. I'm going to call this mainpy. We're also going to open up the command line and we're going to install a package called openai-whisperer. And this is one of those models that you can use without tokens. So, you can just install the package and you can use it without an API key or anything. And what we're going to do here now is we're going to say import whisper. And then we're going to say model equals whisper.loadModel. And we're going to load the base model. And the result of the transcription is going to be just a model.transcribe. And we're going to pass here the video underscore sound.mp3 file. And then we just want to open a new file transcription.txt in writing mode sf f.write. And we're going to write a result into that file, but not the whole result, but just the text that is produced. And that is the whole magic. I think this is running locally. So, we're not using some model in the cloud from OpenAI. I think this installs basically the model locally. I don't know if you have to have certain GPUs or something that are a little bit more modern. I ran this also on my laptop that is now five years old and it has an AMD GPU. So, you definitely don't have to have Nvidia. It worked perfectly fine. Now, let's see if it works here as well. I think we're going to skip the whole process because it takes, oh, actually it doesn't take too long. So, we can actually wait for this. Or was this just a download? I think this was just a download. So, it will probably take some time. So, we can skip to the part where the transcription is done. All right. So, it seems like the transcription is now done. So, we can open up this transcription.txt file and we can see here what's going on, guys. Welcome back. In this video today, we're going to learn about the Python package that allows us to automatically find and so on. So, this is very accurate. This is exactly what I'm saying in the video, but I'm sure we are going to find some mistakes. Modules and packages that are used in a project automatically create a requirements.txt file. Let's get right into it. All right. So, for this video today, I'm going to be working on Linux and so on. I think the package that we used here was pipreqs. So, P-I-P-R-E-Q-S. I think this is definitely not going to be transcribed properly. So, let's see if it's somewhere occurs. Ah, pipreqs. There you go. So, this doesn't work because, of course, pipreqs, the package name P-I-P-R-E-Q-S is not going to be recognized by OpenAI Whisperer because it's just not a commonly used thing. But pipreqs, because reqs is a word and pip is known as a package manager, this is going to be recognized. So, you will have to do some manual adjustments here probably or you can just ignore it. But other than like these special names, I think everything should be transcribed correct. I'm not going to look through all of the text now, but I think that this is a very, very high quality transcription because if you compare this to something like just a basic speech recognition, which you can use. So, you can just use the Python speech recognition module. You can go through the file. You can split it up. You can transcribe the individual pieces. First of all, it's a hassle because you have to not use too much data at once when you transcribe like that. But then again, every single word is probably or every other word is going to be incorrectly transcribed. So, this is a very high quality transcription and all of this is running locally on your machine for free. So, you don't even have to use an API key. You don't have to use tokens or anything. I think you have to have some decent hardware. But as I said, I'm running this now on a pretty strong computer, but I also ran it on my laptop, which is not as strong. It's not even strong enough to record properly. So, I think that most of you guys will be able to run this to some degree. All right. So, that's it for today's video. I hope you enjoyed it and I hope you learned something. If so, let me know by hitting the like button and leaving a comment in the comment section down below. And of course, don't forget to subscribe to this channel and hit the notification bell to not miss a single future video for free. Other than that, thank you much for watching. See you in the next video and bye.