Improved Robustness to Disfluencies in Rnn-Transducer Based Speech Recognition

IEEE International Conference on Acoustics Speech and Signal Processing, pp. 6878–6882

Abstract

Automatic Speech Recognition (ASR) based on Recurrent Neural Network Transducers (RNN-T) is gaining interest in the speech community. We investigate data selection and preparation choices aiming for improved robustness of RNN-T ASR to speech disfluencies with a focus on partial words. For evaluation we use clean data, data with disfluen- cies and a separate dataset with speech affected by stuttering. We show that after including a small amount of data with disfluencies in the training set the recognition accuracy on the tests with disfluencies and stuttering improves. Increasing the amount of training data with disfluencies gives additional gains without degradation on the clean data. We also show that replacing partial words with a dedicated token helps to get even better accuracy on utterances with disfluencies and stutter. The evaluation of our best model shows 22.5% and 16.4% relative WER reduction on those two evaluation sets.

Authors 4

1. Valentin Mendelev

   Amazon (United States)

   Affiliation as printed

   Amazon Alexa

   Amazon

2. Tina Raissi Aachen Human Language Technology and Pattern Recognition Group

   RWTH Aachen University · Amazon (United States)

   Affiliation as printed

   Human Language Technology and Pattern Recognition Group, RWTH Aachen University, Germany

   Amazon

3. Guglielmo Camporese Aachen

   RWTH Aachen University · University of Padua

   Affiliation as printed

   University of Padova, Italy

   RWTH Aachen University

4. Manuel Giollo

   University of Padua · Amazon (United States)

   Affiliation as printed

   Amazon Alexa

   >University of Padua