Speaker 1: Hello everyone, I am Krishna and in today's video, I am going to share with you how you can install Whisper AI in Macbook and how you can use it. So first of all, let's understand what Whisper AI is. So, Whisper AI is an AI that you can use to convert audio to text. What I mean to say is that you can easily convert audio to text. Sorry, I mean audio to text. I apologize for the mistake. So, you can convert audio to text. And you can see here that this is Whisper AI. In this, it says that you have English speech recognition and robustness and accuracy. It means that it has a very accurate practice. And this is ChatGPT, you must have heard of it. ChatGPT is an open source natural net. So, you can use it to easily convert audio to text, which is totally free. So, let's start and see how we can install it and how we can use it in Macbook. And there is a similar process in Windows. So, you can follow this and install it in Windows. But definitely your applications and some processes will be slightly different. So, let's start. So, first of all, we have to install some things. And first of all, we have to install Python. So, let's start Python. So, the Whisper AI is not in support after Python 3.10. There is no latest update. So, now we will install Python 3.10. For this, you have to go to the Python website. I will also share this URL with you. But you have to go to Python. And if I talk from there, let's see on all releases. So, here you can see all releases. But I want Python. 10 point something. So, let me do this. 10.10 is the stable version. So, let me install it. So, for this, you will click on it. Then here, as soon as you scroll down, you can download the installer file according to your operating system. So, in my case, there is Mac OS here. So, I downloaded the file of Mac OS. So, our Python file has been downloaded. Now let's go to the file. And here you can see that I have the .pkg file. I will install it. Continue, continue, and then continue. Obviously, you will have to read the agreement. So, now I have to put my password. And in the same way, you have to follow all the steps. So, sometimes it takes a little time. Basically, what happens is that the package and files will be split all over the place. So, this is the installation time of the file. And now we are installing. Until then, let's understand what we have to install next. So, this is our package. Python is installed. Python is installed. You can also validate it from here. So, IDLE is running here. So, you can see here. And then I try to print. Hello. You can see. So, I said print hello. And here my Python has been printed. This means that my Python has been installed successfully. Now, let's come to the fact that Python has been installed. Now, it is asking me to move the installation file to the bin. I said yes, move it to the bin. I can also validate it. To validate, I started my terminal. And in the terminal, I will type python3- v-v-e-r-s-i-o-n So, you can see the version of Python 3. I got 3.10.10 So, I just downloaded it. So, I have downloaded it. Now, it's our turn. After installing Python, it's my turn to install PyTorch. Now, you will ask what is PyTorch? PyTorch is a machine learning library that is activated on your local system. The machine learning library comes in the process. For that, we use PyTorch. So, when you come here, you have to go to PyTorch and click on get started. And then start locally. So, we are on the PyShort website. Get started and then start locally. So, we come down here. Now, you can select your OS from here. So, now the stable version of PyTorch is 2.2.0. So, I chose this. My machine is Mac. Then I will start using the package pip. Those who are aware of Python, they know what is pip. Pip is a package library. So, I will use the package pip. Here is the Python language. And along with this, there is also a compute platform. According to the system, everyone has a different computer platform. So, in my case, the default will be used. So, now we come to this. I have the installation URL. And you can say that it is ready. So, I will copy it. After copying, we go to the terminal. And in the terminal, we will simply type. Torch Audio. So, here we have typed it. So, I have already installed it. So, basically, it will be installed in your case. And after installing, it is saying that if you want, you can upgrade pip3. But I don't want to upgrade. So, this is how we will install PyTorch. After installing PyTorch, it is our turn. To read the audio file, we have to install ffmpeg. So, basically, this is a library from which the machine can read the audio file. So, to install it, if Homebrew is installed in Mac, then that is easy. Otherwise, you can install Homebrew. So, how do we install Homebrew? So, let's check it once. So, we have the Homebrew website. And here is the installation. So, you don't have to do anything. You will copy it here. And then you can go to the terminal and install Homebrew. Otherwise, because Homebrew is already installed in my system. So, I am not going to share the installation process of Homebrew. Otherwise, you can check the installation process of Homebrew from this URL. So, now it is our turn to use Homebrew to install ffmpeg. So, let's install ffmpeg. So, we have come to the terminal. And we will mention ffmpeg in the terminal. Homebrew install ffmpeg. So, I entered this. And Homebrew will start installing ffmpeg. For this, Homebrew first updates itself. And then the installation process starts. So, I am installing ffmpeg. So, I have successfully installed ffmpeg. Now, after the installation, I will recap once again. First of all, we installed Python. In which 3.10.10 was installed. Which you can say 3.10.10. And after that, we installed PyTorch. In which I used the package pip. And after that, we installed Homebrew. And if Homebrew is already installed, then well. Otherwise, install Homebrew. And after installing Homebrew, we installed AudioReadLibrary. Which is called ffmpeg. I installed it using Homebrew. So, now we have installed it. Now we have to install WhisperAI. So, let's install WhisperAI. For this, we will mention pip3 install And dash capital U. Then open ai-whisper. So, now we install WhisperAI. And to install WhisperAI, we will type pip3 in the command line. Which I told you that pip3 is used by the library of Python. Install dash U. Now you will ask, what is dash U? In case it is already installed in your system, then upgrade it. Then open ai-whisper. So, now we hit enter. Oh, sorry. It's my mistake. I typed the wrong spelling. So, now we have installed open ai-whisper. Now it's our turn to see how to use open ai-whisper for audio text conversion. So, let's see a demo. So, before that demo, let me tell you that Whisper supports many languages. Hindi, English. It supports languages of more than 40 countries. It has three levels. Normal, medium and large. Now, what happens in this is that a lot of attention is paid to accuracy. Like if you use large, then where there is a full stop, where there should be a comma, where what is there. A lot of attention is paid to everything. In the normal level, the whole sentence is converted. But in that, the exact full stop, comma, etc. You will see all this in the medium level. So, let's start. So, let's see how to use open ai-whisper. So, first of all, let's use open ai-whisper. So, to use open ai-whisper, first of all, let's go to the location of the file. So, in my case, I have kept the file on the desktop. And I have kept the file in a folder called audio. So, I went to the audio folder of the desktop. After that, you just have to use whisper. h-i-s-p-e-r Whisper and then quote. And then the name of the audio file. So, I have also kept the name of the file as audio.mp3. With extension. Okay. And just you can simply enter it. So, this is by default, it will analyze the language. What is the language of this audio? And after that, it will analyze that if you have not defined any level. So, this is generally the lowest level, which we call. It will start executing on it. Okay. So, whisper is now running. I am trying to activate. So, you can see here that he said that it does not support you. So, it started using something like pf32 machine type. And you can see here that the detected language is English. And here is my file has been converted. In which it was told that from 0 to 8 seconds. Hello everyone, Krishna is there. So, first of all, I tell you the audio file. Okay. So, this is my audio file. And let's play it. Hello everyone, Krishna here. And today I am going to share you how to install Whisper AI. Okay. Notes. Let's go to the notes first. So, now you can see that I have converted the text file. Hello everyone, Krishna here. And today I am going to show you how to install Whisper AI. So, now you can see that it has been converted to the full JSON format. If you want to see the simple text, then you can open the simple text file. And it will be completely text only. So, you can see how simply we have converted an audio file with accuracy as a text. Okay. So, now let's go back to the Whisper in use cases. So, how can we do use cases? So, in use cases, you have Whisper. There are many methods in which you can run the model. Tiny, Base, Lower, Medium, Large. So, generally it runs in the lower case. But you can define the model. So, in my case, I will define the model as Medium. So, in this case, what will happen is that your Medium. But definitely as you increase the size of the model, So, in that case, your RAM or GPU consumption will be more. So, in that case, you have to keep in mind that you have a higher GPU size. That is, you have a GPU of 4GB or 5GB or you have a high RAM capacity. Okay. Yeah. Okay.