Speaker 1: Welcome to Locals Only Open Source AI, the place to go for the latest open source, absolutely free, artificial intelligence solutions. Today, we're going to dive into something exciting, installing OpenAI's Whisper. Are you tired of paying for audio and video transcription? Do a lot of open source installation videos leave you lost, frustrated, or even more confused? Then join me as I navigate the vast, ever-growing world of artificial intelligence. Remember, I am not a professional, just an average human searching for new and cost-effective ways to tackle complex tasks by using AI. Whisper is an automatic speech recognition system developed by OpenAI that supports multiple languages. Running this model locally is simple. All you need is a few quick installations and your browser. In this tutorial, I'll walk you through how to install and run the model locally on your computer. Let's get started. First, you'll need to clone the repository, open a new browser, and go to github.com forward slash OpenAI forward slash Whisper, and then press enter. This will take you to the OpenAI Whisper main page, where you will download the repository for installation. This page contains a lot of important content, including setup information, so I do encourage you to take a few minutes and scan the information. When you are ready, select the code button and then select download zip. I have a directory created on my external hard drive, so I will create my OpenAI Whisper folder and download the repository there. Once the download is complete, extract the contents to your new Whisper folder. I am using 7-Zip. Now let's navigate to the WhisperAI download folder and view the newly extracted zip file. I'm going to move all of my files to the root WhisperAI folder that I created and delete the empty folders. This is a personal preference, and not necessary. Now inside your root folder, in the address bar of Windows Explorer, enter CMD and then press enter. Next, we need to copy the WhisperWeb GPU repository to your WhisperAI folder. Once you have copied the repository, you will need to go to the WhisperAI folder and then press enter. Next, we need to copy the WhisperWeb GPU repository to your WhisperAI folder. Enter the command located on the bottom of your screen into the command window and then press enter. Once that process has complete and your command prompt has returned, we need to install a few setup tools. First, we will install Rust by entering the command located at the bottom of your screen into the command window and pressing enter. Next, we will install FFmpeg by entering the command located at the bottom of your screen into the command window and pressing enter. As you can see, my system already has this installed. Now let's change to the whisper-web directory by entering the command located at the bottom of your screen into the command window and pressing enter. Next, we need to download Node.js and install it. In your address bar, enter the link located at the bottom of your screen into your browser's address bar and press enter. Copy the script to your clipboard, return to your command prompt, and then press ctrl-v. Select paste anyways when the dialog appears and then press enter. Let's check our node version by entering node-v. Now let's install npm by entering the command located at the bottom of your screen into the command window and pressing enter. If you receive a message indicating that packages are looking for funding, enter the npm fund command located at the bottom of your screen into the command bar and then press enter. We're almost finished. In the command window, enter the npm audit fix command located at the bottom of your screen into the command bar and press enter. Let's just verify the install by entering the npm-v command located at the bottom of your screen into the command bar and pressing enter. That's it. We are now ready to start the server by entering the npm run dev command located at the bottom of your screen into the command window and pressing enter. The app uses web gpu technology, meaning the model runs entirely on your device. You can upload an audio file, load it, and once the model is loaded, transcription will begin. Remember, performance may vary depending on your computer. Once complete, you can download the transcription in text or JSON format, complete with timestamps and punctuation. Copy the local host address and then paste it into your web browser. On the whisper graphic user interface page, you have three options to choose from. The first option allows you to import a file from a url. As you can see, by default, a sample selection has been pre-loaded for you. Select load and then select play. Then select transcribe to begin the transcription. You can export the transcription as either a text or JSON file. The second method allows you to upload a locally saved file. We have selected a random music file for demonstration purposes. The third and final method allows you to use a microphone record your transcription text. The app also works with video files. If you encounter any issues while transcribing, just refresh the page. Additionally, you can record your voice and convert it to text. And that wraps it up for this video. I have included all of the commands and links used in this video in the description below. As always, if you found this video helpful, don't forget to leave a comment, hit the like button, and subscribe to Locals Only Open Source AI to help me grow the channel and make new videos. Thanks for watching, and I'll see you in the next video.