Browse Prior Art Database

Automatically Reconfiguring Error Correction Method for Speech Communications System

IP.com Disclosure Number: IPCOM000043255D
Original Publication Date: 1984-Aug-01
Included in the Prior Art Database: 2005-Feb-04
Document File: 3 page(s) / 62K

Publishing Venue

IBM

Related People

Irvin, DR: AUTHOR

Abstract

This article describes a self-configuring error correction method for use in a multi-user digital speech communications system. Errors occurring on a digital communications channel are detected, and the degree of error protection applied through forward error correcting (FEC) code is adjusted to match the error characteristics of the channel. Transmission of digital data requires greater protection than transmission of speech since the effect of noise degradation on digital data interpretation is much greater than the perceived effect of distortion in speech. The approach chosen is one of taking extra bits from the speech channel allocation and reallocating them to error correcting code for the digital data portion of the traffic. As shown in U. S.

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 48% of the total text.

Page 1 of 3

Automatically Reconfiguring Error Correction Method for Speech Communications System

This article describes a self-configuring error correction method for use in a multi-user digital speech communications system. Errors occurring on a digital communications channel are detected, and the degree of error protection applied through forward error correcting (FEC) code is adjusted to match the error characteristics of the channel. Transmission of digital data requires greater protection than transmission of speech since the effect of noise degradation on digital data interpretation is much greater than the perceived effect of distortion in speech. The approach chosen is one of taking extra bits from the speech channel allocation and reallocating them to error correcting code for the digital data portion of the traffic. As shown in U. S. Patent 4,357,700, for example, communications channels experiencing variable error rates may be reconfigured in response to the perceived bit error rate by adding forward error correcting codes or assigning additional channel bandwidth for these expanded code streams to preserve the integrity of the transmitted data. In speech communications systems where data is transmitted as a secondary function, and in particular for such systems in which a high degree of statistical compression in the voice spectrum has been employed, the present technique for providing the additional capacity to preserve the digital data gracefully trades fidelity in the speech portion for error protection of the data. Fig. 1 illustrates the proposed technique as applied to a ten-user speech communications system. Each voice source or user is digitally encoded by, for example, a 16 K-bps sub-band speech coder. This results in 15-millisecond data blocks with 240 bits per block. Two hundred bits are assigned to the speech encoding function and 40 bits are for the bit allocating signalling function to describe how the bits in the 200-bit segment have been allocated. In the example shown, 13.34 kilobits per second per coder are devoted to the transmission of speech and 2.66 kilobits per second bit allocations are given for signalling under normal operating conditions. This is called, for later reference purposes, the state 1 condition. When the signalling information is altered due to channel errors, proper reassembly of speech bits is not possible, but the perceived effect is not so significant as it is where data bits are being reassembled. The signalling information then must be protected from channel errors as the allocation of the speech bits will be lost and fidelity will be destroyed. It is the signalling information then that must be highly protected either for the protection of following data or for maintenance of some audible fidelity. Traditional error correcting codes, such as FEC codes, require substantial overhead in the number of bits available for coding the desired speech. This means that the remaining bandwidth avail...