Patent · US Active

Channel-compensated low-level features for speaker recognition

US10347256B2 · kind B2 · utility

16Cited by
32References
28Claims
0Family size

Assignee

Inventors

Key dates

Filing dateSep 19, 2017
Grant dateJul 9, 2019
Priority date
Expiry dateJan 16, 2038

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG10L19/028
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

A system for generating channel-compensated features of a speech signal includes a channel noise simulator that degrades the speech signal, a feed forward convolutional neural network (CNN) that generates channel-compensated features of the degraded speech signal, and a loss function that computes a difference between the channel-compensated features and handcrafted features for the same raw speech signal. Each loss result may be used to update connection weights of the CNN until a predetermined threshold loss is satisfied, and the CNN may be used as a front-end for a deep neural network (DNN) for speaker recognition/verification. The DNN may include convolutional layers, a bottleneck features layer, multiple fully-connected layers and an output layer. The bottleneck features may be used to update connection weights of the convolutional layers, and dropout may be applied to the convolutional layers.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.