• Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint
Share this Page URL
Help

Chapter 10. Speech Coding > Overview of MPEG-4 Speech Coders

10.2. Overview of MPEG-4 Speech Coders

MPEG-4 speech coding tools [MPEG4-3] consist of two different basic algorithms: CELP [Nomu98] and HVXC [Nish97a]. They provide not only high coding efficiency but also a variety of new functionalities, such as bit-rate scalability (also known as embedded coding); bit-rate controllability, which allows modification of the bit rate during encoding; variable bit-rate coding; and speed and pitch changes. Users can choose the best algorithm for their applications and requirements.

The CELP algorithm operates from 3.85 to 23.8 kbit/s for speech signals of 8- and 16-kHz sampling rate with two different excitation modes—that is, multipulse excitation (MPE) and regular pulse excitation (RPE). Parametric speech coding, HVXC, fits for even lower bit-rate coding of speech signals at 8-kHz sampling rate. HVXC uses a parametric representation of the harmonic spectral magnitudes of LPC residual signals for voiced segments, and a vector excitation coding (VXC) algorithm for unvoiced segments. Bit-rate scalable coding is enabled by a multistage structure of the vector quantization (VQ) scheme for LPC parameters and LPC residuals. In addition, speed and pitch changes are possible during decoding by manipulating the encoded parameters due to the parametric representation of the speech signals.


PREVIEW

                                                                          

Not a subscriber?

Start A Free Trial


  
  • Creative Edge
  • Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint