6,138,092 A * 10/2000 Zinseretal 704/223
6,233,550 Bl 5/2001 Gershoetal.
6,308,155 Bl* 10/2001 Kingsbury et al 704/256.1
6,453,283 Bl * 9/2002 Gigi 704/207
6,675,144 Bl* 1/2004 Tucker et al 704/264
6,681,202 Bl* 1/2004 Mietetal 704/214
6,940,454 B2* 9/2005 Paetzold et al 345/473
6,985,857 B2 * 1/2006 Adut 704/230
7,283,585 B2 * 10/2007 LeBlancetal 375/222
2002/0016698 Al* 2/2002 Tokuda 702/190
2002/0052738 Al * 5/2002 Paksoyetal 704/219
2003/0050786 Al * 3/2003 Jax et al 704/500
2004/0138874 Al * 7/2004 Kaajasetal 704/205
2004/0153544 Al * 8/2004 Kelliheretal 709/226
2005/0055219 Al* 3/2005 Accardi et al 704/278
2006/0153286 Al * 7/2006 Andersen et al 375/229
2007/0110042 Al* 5/2007 Lietal 370/352
OTHER PUBLICATIONS
Jacovo et al., Some experiments of 7 kHz audio coding at 16 kbit/s, IEEEICASSP 1989, 192-195.*
* cited by examiner
Primary Examiner—David R Hudspeth
Assistant Examiner—Jakieda R Jackson
(74) Attorney, Agent, or Firm—Farjami & Farjami LLP
(57) ABSTRACT
An approach for improving quality of synthesized speech is presented. The input speech or residual is first separated into a voiced portion and a noise portion. The voice portion is coded using CELP methods. The noise portion of the input speech may be estimated at the decoder since it contains minimal voiced speech components. The separation is frequency dependent and is adaptive to the input speech. The separation may be accomplished using a lowpass/highpass filter combination. The information regarding bandwidth of the lowpass/highpass is presented to the decoder to facilitate reproduction of the noise portion of the speech.
44 Claims, 3 Drawing Sheets
