Speech Classification

Challenge

 

A question arised to authenticate a user’s gender for more personalised experience in application. But what can be the natural identifier of a user’s gender ? and at the same time be it spoof proof so someone with false gender identity cant bypass. Visual solutions were off the table as people don’t tend to be comfortable in sharing their pictures.

Solution

 

Speech !. Yes, a user’s voice contains their gender characteristics, so we use the user’s voice to identify the genders. To fool proof this system we further take our RnD up a notch and convert the speech to text too. Now we show the user a text phrase and the user says it. To complete the process the phrase shown and the phrase said need to be similar.

Technologies Used

Speech Classification Uses

  1. Profiling
    • These systems can be expanded to detect a user’s age region or dialects.
  2. Automated home
    • Send commands as switched to control a home.
  3. Automated orders
    • Drive through restaurants can process orders quickly.
  1. Filling medical prescription
    • Doctors’ prescriptions can be automated as they consult a patient.
  2. Authentication
    • Various security authentications can use a voice authentication system for ease of access.
  3. Sound surveillance
    • Other than human voice it can be used for surveillance of other sounds for scientific purposes.