Pobierz Notatka przedstawiająca działanie Whisper AI i więcej Notatki w PDF z Badania jakościowe tylko na Docsity! 1. Go to the website https://colab.research.google.com, i.e. the space for managing virtual notebooks. You need to be logged in to your Google account. 2. Select File and then New Notebook. It will be saved to your Google Disk. 3. The notebook is named Untitled1.ipynb (or similar, with different numbering) by default. Change the name to one that is short and unambiguous so that you can easily find it in the future, e.g. TranscriptionsWhisperAI.ipynb. The .ipynb file extension stands for Python code notebooks. 4. Use GPU as an accelerator. Enable the GPU setting by Edit -> Notebook Settings and choose T4 GPU. Otherwise, your transcription will not start and will return errors! 5. Click on the space with the Play button: 6. Paste the following and click the Play button: !pip install git+https://github.com/openai/whisper.git !sudo apt update && sudo apt install ffmpeg The first line of code will install Whisper AI in the virtual space on your Disk (and NOT physically on your computer). The other line either installs or updates the codec responsible for voice recognition. 7. Click on the Folder button on the bottom left: It will show your workspace where you upload your files. Drag them into this space. Alternatively, you can enter the following code (first click Code above, a new code entry space will appear) and click Play: from google.colab import files uploaded = files.upload() It will create an action button to open a window to upload files. However, this is, in most cases, entirely optional. In both cases, the files should appear on the left-hand side. If you right-click, you can rename them, for example. Their names should be short and unambiguous to make the files easier to manage. It is important because the output files will be named based on the input files, i.e. those with recordings you want to transcribe.