Published on Sat Jul 06 2019

Towards Debugging Deep Neural Networks by Generating Speech Utterances

Bilal Soomro, Anssi Kanervisto, Trung Ngo Trong, Ville Hautamäki

Deep neural networks (DNNs) are able to successfully process and classify speech utterances. However, understanding the reason behind a classification by DNN is difficult. One such method used with image classification DNNs is activation maximization.

0
0
0
Abstract

Deep neural networks (DNN) are able to successfully process and classify speech utterances. However, understanding the reason behind a classification by DNN is difficult. One such debugging method used with image classification DNNs is activation maximization, which generates example-images that are classified as one of the classes. In this work, we evaluate applicability of this method to speech utterance classifiers as the means to understanding what DNN "listens to". We trained a classifier using the speech command corpus and then use activation maximization to pull samples from the trained model. Then we synthesize audio from features using WaveNet vocoder for subjective analysis. We measure the quality of generated samples by objective measurements and crowd-sourced human evaluations. Results show that when combined with the prior of natural speech, activation maximization can be used to generate examples of different classes. Based on these results, activation maximization can be used to start opening up the DNN black-box in speech tasks.

Sat Mar 10 2018
Machine Learning
Speech Recognition: Keyword Spotting Through Image Recognition
The problem of identifying voice commands has always been a challenge. We will compare the efficacies of several neural network architectures for the speech recognition problem. The result is a demonstration of how to convert a problem in audio recognition to the better-studied domain of image classification.
0
0
0
Tue Jan 02 2018
NLP
Did you hear that? Adversarial Examples Against Automatic Speech Recognition
Researchers demonstrate powerful attacks against machine learning models that can fool them to produceincorrect results. Algorithm performs 87% success by adding small background noise without knowing the underlying model parameter and architecture.
0
0
0
Sat Sep 14 2019
Machine Learning
NeMo: a toolkit for building AI applications using Neural Modules
NeMo (Neural Modules) is a Python framework-agnostic toolkit for creating AI applications. NeMo is built around neural modules, conceptual blocks of neural networks that take typed input and produce typed outputs.
0
0
0
Wed Feb 19 2020
Machine Learning
Gradient-Adjusted Neuron Activation Profiles for Comprehensive Introspection of Convolutional Speech Recognition Models
0
0
0
Wed Jun 09 2021
NLP
Unsupervised Automatic Speech Recognition: A Review
Automatic Speech Recognition (ASR) systems can be trained to achieve remarkable performance given large amounts of manually transcribed speech. Large labeled data sets can be difficult or expensive to acquire for all languages of interest.
2
0
1
Fri Jul 30 2021
NLP
USC: An Open-Source Uzbek Speech Corpus and Initial Speech Recognition Experiments
The Uzbek speech corpus (USC) comprises 958 different speakers with a total of 105 hours of transcribed audio recordings. To ensure high quality, the USC has been manually checked by native speakers. The experimental results demonstrate promising results for the applicability of the USC for ASR.
0
0
0