Browse Prior Art Database

HAND-OVER ENHANCED AUDIO RECOVERY (HEAR) METHOD

IP.com Disclosure Number: IPCOM000008135D
Original Publication Date: 1997-Jun-01
Included in the Prior Art Database: 2002-May-21
Document File: 3 page(s) / 125K

Publishing Venue

Motorola

Related People

Martin T. Karanja: AUTHOR [+3]

Abstract

With the rapid deployment of digital technology in wireless communications, the field of speech coding has seen a tremendous amount of develop- ment. Low bit-rate coding schemes such as CELP and VSFLP have gained a lot of popularity in digi- tal wireless telephony. The basis of these parametric vocoders ("voice coders") is at the encoder speech is modeled as a time-varying vocal tract filter excit- ed by a stochastic excitation sequence. At the decoder, this vocal tract filter/excitation model is reconstructed and the decoded speech samples are recovered. In this vocal tract filter/excitation coding process the output of the encoder depends not only on the present input but also on past inputs. Similarly the decoder output depends on present and past inputs. Furthermore, it is essential that the decoder states track the encoder states.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 50% of the total text.

Page 1 of 3

8 MOTOROLA Technical Developments

HAND-OVER ENHANCED AUDIO RECOVERY (HEAR) METHOD

by Martin T. Karanja, Sedny S.J. Attia and Anthony Cl. Tran

THE PROBLEM

  With the rapid deployment of digital technology in wireless communications, the field of speech coding has seen a tremendous amount of develop- ment. Low bit-rate coding schemes such as CELP and VSFLP have gained a lot of popularity in digi- tal wireless telephony. The basis of these parametric vocoders ("voice coders") is at the encoder speech is modeled as a time-varying vocal tract filter excit- ed by a stochastic excitation sequence. At the decoder, this vocal tract filter/excitation model is reconstructed and the decoded speech samples are recovered. In this vocal tract filter/excitation coding process the output of the encoder depends not only on the present input but also on past inputs. Similarly the decoder output depends on present and past inputs. Furthermore, it is essential that the decoder states track the encoder states.

  In a cellular environment, as a mobile sub- scriber unit (SU) traverses cell boundaries, it will go from one carrier frequency to another. This con- cept is known as "hand-over." There are two call scenarios in cellular telephony: mobile-to-mobile and mobile-to-landline. In a mobile-to-mobile call where SUl is the subscriber unit undergoing hand-over and SU2 is not undergoing hand-over,

all vocoder activity in SUl is suspended during hand-over. This will result in a mute in the output audio of both subscriber units. Upon recovery from hand-over SUI will reset its vocoder. However the vocoder in SU2 will continue along its respective processing state (non-reset state). Similarly in the mobile-to-landline case the mobile will reset its vocoder but the fixed network equipment (FNE)

will not reset the transcoder. This vocoder state mismatch can result in audio degradation upon recovery from hand-over until the state of the decoder in SUl re-synchronizes with the encoder in SU2 (or the transcoder). Depending on the nature of the speech being encoded at the time of hand-over, the duration of audio degradation can be as long as one second.

THE SOLUTION

  Since the problem is a mismatch in the vocoder states at each digital end-point (SU or transcoder), the solution is to reset both vocoders upon comple- tion of the hand-over process. The proposed Hand- over Enltanced Audio Recovery (HEAR) method is to introduce a signalling scheme between the two digital end-points instructing the SU or tmnscoder to reset its vocoder. The HEAR method is descr...