Representing Nonspeech Audio Signals through Speech Classification Models

Huy Phan, Lars Hertel, Marco Maass, Radoslaw Mazur, Alfred Mertins


The human auditory system is very well matched to both human speech and environmental sounds. Therefore, the question arises whether human speech material may provide useful information for training systems for analyzing nonspeech audio signals, such as in a recognition task. To find out how similar nonspeech signals are to speech, we measure the closeness between target nonspeech signals and different basis speech categories via a speech classification model. The speech similarities are finally employed as a descriptor to represent the target signal. We further show that a better descriptor can be obtained by learning to organize the speech categories hierarchically with a tree structure. We conduct experiments for the audio event analysis application by using speech words from the TIMIT dataset to learn the descriptors for the audio events of the Freiburg-106 dataset. Our results on the event recognition task outperform those achieved by the best system even though a simple linear classifier is used. Furthermore, integrating the learned descriptors as an additional source leads to improved performance.
Original languageEnglish
Title of host publicationProc. 16th Annual Conference of the International Speech Communication Association (INTERSPEECH 2015)
Number of pages5
Place of PublicationDresden, Germany
PublisherInternational Speech and Communication Association (ISCA)
Publication date01.09.2015
Publication statusPublished - 01.09.2015
Event16th Annual Conference of the International Speech Communication Association - International Congress Center Dresden, Dresden, Germany
Duration: 06.09.201510.09.2015
Conference number: 118697


Dive into the research topics of 'Representing Nonspeech Audio Signals through Speech Classification Models'. Together they form a unique fingerprint.

Cite this