Speech and handwriting

  • András Kornai
Part of the Advanced Information and Knowledge Processing book series (AI&KP)

Conceptually, the techniques of linguistic pattern recognition are largely independent of the medium, but overall performance is influenced by the preprocessing to such an extent that until a few years ago the pattern recognition step was generally viewed as a small appendix to the main body of signal processing knowledge. To this day, it remains impossible to build a serious system without paying close attention to preprocessing, and deep algorithmic work on the recognizer will often yield smaller gains than seemingly more superficial changes to the front end. In Section 9.1, we introduce a speech coding method, linear prediction, that has played an important role in practical application since the 1970s.We extend the discussion of quantization started in Section 8.1 from scalars to vectors and discuss the Fourier transform-based (homomorphic) techniques that currently dominate the field.


Speech Signal Discrete Fourier Transform Speech Perception Vector Quantizer Vocal Tract 
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-Verlag Berlin Heidelberg 2008

Authors and Affiliations

  • András Kornai
    • 1
  1. 1.MetaCarta Inc.CambridgeUSA

Personalised recommendations