Where can I find the tutorial, how to train the Azure Video Indexer using our own datasets and tags?

Question

I want to use Azure Video Indexer to train our own video datasets to make it recognize some key words and action in our own dataset. I do not find where I can train Azure Video Indexer in tutorial? How can I train our datasets on Video Indexer?

Answer 1

Thanks for the details, To train Azure Video Indexer with your own custom video data, you can use the customization features provided by Azure Video Indexer. Here are the steps you can follow:

https://learn.microsoft.com/en-us/azure/azure-video-indexer/customize-speech-model-overview

1. Customize a Speech Model: Azure Video Indexer allows you to customize a speech model to improve the transcription of your videos. You can train the model with datasets that include plain text data and pronunciation data.
2. Customize a Language Model: You can customize the language model by uploading adaptation text, which is text from the domain whose vocabulary you’d like the engine to adapt to. Once you train your model, new words appearing in the adaptation text will be recognized, and the language model will learn new probable sequences of words.
3. Edit the Transcript: By editing and correcting the transcript generated by Azure Video Indexer, you’re training a language model to provide improved results in the future.
4. Tagging and Descriptions: After the video has been indexed, you can manually add your own tags and descriptions to the video on the Video Indexer portal. This will help in improving the searchability of your videos.