Speaker 1: If you have tried to transcribe and generate subtitles for your videos, you will know that it is a slow and exhausting process. Even the automatic transcription of YouTube requires going sentence by sentence, correcting technical words, tool names, adding capital letters, punctuation marks, spelling, etc., etc. But the OpenAI Whisper AI tool that I show you below greatly speeds up this task given its high precision. This audio-to-text transcriber is open source and achieves human precision in recognizing a wide variety of languages, including Spanish, where it records an error rate of less than 5%. In the latest version 3, I barely found errors and was able to make transcriptions and generate .srt files in a matter of minutes. If you want to get the same results, don't miss this video. I explain how you can use Whisper, how to subtitle and transcribe your videos, how to upload the subtitles to your YouTube channel, how to upload the subtitles to Vimeo, and also an alternative for Mac users. Hi, I'm Merchel Viñas. I help teachers, trainers, and students learn and teach with new technologies, including artificial intelligence. If this is the first time you visit the channel, consider subscribing by clicking on the button below. My videos will help you transform your classes and advance professionally. You will find links to all the tools I mention in the video below in the description. Whisper has been trained with 680,000 hours of data from multilingual audio extracted from the web. This extensive training allows you to better recognize accents, background noises, and technical terms. In addition, Whisper goes further and facilitates transcription in several languages, allowing translation into English. As the only drawback, the length of each segment in the subtitles is longer than I would like, and cutting them would again be an extensive manual process. The use of this tool has many educational advantages. For example, converting educational videos into text for review and accessibility. Also, automatically capturing face-to-face classes to generate detailed notes or simplifying taking notes. Using transcriptions to learn languages, improving vocabulary and pronunciation. Transcribing interviews and debates for review and analysis of ideas expressed in group sessions. And also transcribing events, conferences, and training workshops. To be able to use Whisper and transcribe your videos or third parties, you should know that Whisper does not have a web page to download the software directly to your computer. Although all files are available on GitHub, an online platform to share code, installing these files on your PC can be a technical and complex process, not suitable for any teacher. However, there are several options on the Internet that host the tool online, such as Replicate and Hug & Face. If you are a Mac user, there is also an application, Mac Whisper, which I show at the end of this video and allows you a higher level of privacy when transcribing on your PC and not on the web. To transcribe an audio file from Replicate, click on the following link, you will find it below in the description. Upload an audio file in mp3, mp4 or WAV format so that the page can recognize it. There is no limit on the file duration. You will see that it uses the version of the LART V3 model. The transcription is slower than with lower models, but it is also the most accurate in recognition of speech. Choose the .srt format if you want to get subtitles with timestamps or simply plain text if you want it to be the written text of the video. Within the language, in the drop-down menu, choose Spanish or the language in which the audio of your file is, although it also detects it without specifying it. Keep going down and click on the black button that says Run. Now, on the right of the screen, the transcription starts, marking the progress of the transcription here in percentage. When you have finished the transcription, select the text with the mouse, copy it and paste it in a document. Review the text and correct any error. It is usually perfect, but check at the end of the video since it sometimes adds the phrase, thanks for watching the video, and simply delete this phrase. If you need subtitles, you have to create a .srt file and for this, paste the text in the Windows Notes block or in the TextEdit de Mac. In the TextEdit de Mac, hold down the Option key and then click on File, Save As and give it the .rst termination. You have already created the subtitle file to upload it to YouTube. To subtitle a YouTube video, log in to your YouTube Studio account, select the left tab, Subtitles, choose the video to subtitle, add the language of the file, click on Add under the Subtitles column and upload the .srt file from your PC. To upload the .srt file and create the subtitles in Vimeo, log in to your library and open the subtitle video. Select the CC, Legends and Audio button on the right of the player, click on Hidden Subtitles, click on the plus button on the right of My Uploads, then specify the language and file type. Click on Choose a file and upload the subtitle file from your PC. Once the file is uploaded, you will have to activate the subtitles and the CC symbol will appear on the player. You will also see a transcription of the text, timed with the video, which is also very useful, it allows you to search to jump to a specific point in the video. If you are a Mac user, you have the option of installing a software called Mac Whisper for a single cost of 19 euros. The content is transcribed locally, obtaining greater privacy of the content and can be done without internet connection, accelerating the transcription and creation of the .srt file. It is the program that I am currently using for the transcription of the videos of my online courses. The transcription consumes a large amount of CPU, so having the M1 or M2 chip on the Mac is recommended. I have installed the large model, the V3, on my Mac, which occupies about 3 GB, but it allows you to have greater precision, but you can also install smaller models of previous versions. For you to get an idea, a perfect transcription of 5 minutes of audio took 1 minute and 50 seconds to complete. Mac Whisper also recognizes several people speaking in an audio, facilitating the transcription of interviews or face-to-face classes with various tutorials. And another interesting option is that you can paste the web link of a YouTube video to transcribe it automatically. If you need to create subtitles with Mac Whisper, you will only have to click on Export and choose the .srt or .vtp option according to the video platform you use. If you want to be informed about educational technology or artificial intelligence applied to education to be more productive, don't forget to subscribe to the channel by clicking on the link below. You can also keep learning with the videos you see now on the screen. Until next time.