×

Method and apparatus for diphone aliasing

  • US 6,122,616 A
  • Filed: 07/03/1996
  • Issued: 09/19/2000
  • Est. Priority Date: 01/21/1993
  • Status: Expired due to Term
First Claim
Patent Images

1. A method for speech synthesis in an electronic speech synthesis system, the speech synthesis method comprising:

  • a) storing in a memory of the electronic speech synthesis system a voice table comprised of a set of phonetic waveforms, each phonetic waveform of the set of phonetic waveforms corresponding to a demi-diphone of the voice table;

    b) receiving as an input to the electronic speech synthesis system a phonetic string representative of speech to be synthesized by electronic speech system, the phonetic string comprising diphones, the diphones comprising demi-diphones;

    c) generating synthetic speech of the phonetic string representative of speech in the electronic speech synthesis system by outputting stored voice table phonetic waveforms by;

    i) retrieving a stored voice table phonetic waveform corresponding to a demi-diphone of the input phonetic string representative of speech in the case of the demi-diphone of the phonetic string representative of speech having a phonetic waveform in the voice table corresponding to the demi-diphone;

    ii) retrieving a stored voice table phonetic waveform not corresponding to a demi-diphone of the input phonetic string representative of speech in the case of the demi-diphone of the phonetic string representative of speech not having a phonetic waveform in the voice table corresponding to the demi-diphone by locating a substitute demi-diphone of the voice table having a corresponding stored voice table phonetic waveform which has phonetic features meeting;

    A) a threshold set of phonetic features of the demi-diphone not having a corresponding stored voice table phonetic waveform, wherein the threshold set describes a minimum set of characteristics that must be share d by;

    1) the demi-diphone of the phonetic string representative of speech not having a phonetic waveform in the voice table corresponding to the demi-diphone and,2) the substitute demi-diphone of the voice table having a corresponding stored voice table phonetic waveform; and

    ,B) the most features in common with the demi-diphone not having a corresponding stored voice table phonetic waveform.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×