Azure ML-Training Model on voice RRS feed

  • Question

  • Is it possible to train a Azure model on Voice?

    I am working on a call center use case where in , when a customer calls in with a complaint based on the description the ML model should be able to predict & classify the Customer Complaint

    Actually I was able to train "two class boosted decision tree" model, on text based complaints using feature hashing technique, but specifically looking to train the model on Voice

    Guidance, examples & references would help

    Incase OOB feature wouldn't exist in MLS, open to hear about workarounds as well


    Monday, January 27, 2020 9:44 PM

All replies

  • Hi,

    Can you please add more details about the input data and classes that you want to predict and classify? We have forwarded this feedback to product team for best practices.

    This github sample will help you to build a NLP that can classify a given question into a category. You can clone this git into a Notebook VM in AML Workspace and run ‘bert-stack-overflow/1-Training/AzureServiceClassifier_Training.ipynb’ to build a model.


    Wednesday, January 29, 2020 5:32 AM
  • Hi, Thanks for the response. This is the core idea

    Call center would have data Comprising of Label Customer Complaint and Feature Issue description  

    Sample "Customer Complaint" would be Billing issue or Internet speed

    Sample description would be "Hey i was paying $x but the price went up by $y, can you apply some promotion to being my monthly bill down by $z"

    the webservice will take Voice input Issue Description & should product either voice/text output Customer complaint

    We would want to train ML model with the data and deploy as web service. THis way when the consumer calls with an issue, they need not dial options 1,2,3 rather the model is able to route them to appropriate queue

    (1) How do I convert text data to Voice in Azure MLS?

    (2) How do i apply , train & Score model in Voice 

    (3) How to make the webservice accept Voice input to produce the scored label/value as webservice output 

    Wednesday, January 29, 2020 5:36 PM
  • Hi,

    Currently there is n't any notebook samples  available that can handle voice input. You can raise a user voice request here so the community can vote and provide their feedback, the product team then checks this feedback and implements the feature in future releases.


    Friday, January 31, 2020 6:25 AM