What we do
All voice biometrics being marketed today are based upon machine learning techniques. This means a great deal of training data (recordings of voice samples) are required to make the technology work for a given application. That’s why the customer gets only a limited domain application and typically numerous speech scientists are required to train and tune the technology for the limited domain techniques to achieve acceptable accuracy levels.
At MY VOICE we focus on human based auditory modelling of the human voice. The resultant approach is ‘universal domain’ and does not require extensive tuning. The reason human based auditory modelling was chosen is that the human ear is able to identify a speaker’s voice based on anything the speaker says without extensive training or a long learning process. We have made a break-through in universal voice biometrics because its artificial intelligence approach faithfully models the way the human auditory system identifies or recognises a given speaker’s voice independent of language being spoken.