Analysis Techniques for Speech Signals

Part of the Lecture Notes in Electrical Engineering book series (LNEE, volume 13)

In this chapter we will introduce several well known approaches for speech analysis. These approaches include parametric representations of the spectral envelope of a speech signal as well as scalar speech features. These features will later on be used for representing the spectral envelope for classification and extension schemes. The scalar features are interesting for more robust classification as they are able for example to give information on the gender of the speaker or on voicedness of the utterance. The presented methods are related to or originate from the field of speech coding or speech recognition [Mitra 93, Kleijn 95]. To evaluate speech coders some distance measures have been introduced that will also be presented and complemented by a novel measure. These distance measures will allow us to evaluate the quality of the bandwidth extension approaches that will be presented in Chap. 4 and Chap. 5. The focus of this chapter is placed on the linear predictive analysis of speech signals, which will result in a parametric representation of the spectral envelope, since all algorithms that are presented in this book make use of this method.


Discrete Cosine Transform Speech Signal Vocal Tract Parametric Representation Speech Segment 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer Science+Business Media, LLC 2008

Personalised recommendations