Train model Speech to text

Question

Hello,

I did an application that use speech cognitive service to transcript audio files.

I use Speech to Text API v3.0 ( https://westus.dev.cognitive.microsoft.com/docs/services/speech-to-text-api-v3-0/operations/CopyModelToSubscription ) and the model by default.

Now, I would like improve my model, I saw that there is a custom model to create a model with a dataset. But this way it's not really pratice for me.
Because users will correct the text in my program and I want start a automatic train without use the interface from custom model.

Is it possible ?

I saw it's possible to create a dataset and a model with the Speech to Text API v3.0, but I don't find the service to train the model.

I hope my request is clear.

Thanks.

Answer 1

@Amen Lise Thanks for the question. Here is the document to train and evaluate model.

transformer transcribes raw audio files and live audio streams into text. Transcription supports realtime audio streaming, automatic splitting into utterances, and profanity detection. Supports several languages and Custom Speech Models.

Here is link to the custom speech models and how to use custom model.