Method for protecting multi-pulse coders from fading and random pattern bit errors

• US 5,073,940 A
• Filed: 11/24/1989
• Issued: 12/17/1991
• Est. Priority Date: 11/24/1989
• Status: Expired due to Term

• Current Assignee: Ericsson, Inc. (Telefonaktiebolaget LM Ericsson)
• Original Assignee: General Electric Company
• Inventors: Koch, Steven R., Zinser, Richard L., Toy, Raymond L.
• Primary Examiner(s): KEMENY, EMANUEL
• Application Number:
  US07/441,022
• Time in Patent Office: 753 Days
• Field of Search: 381/38, 381/47
• US Class Current: 704/226
• CPC Class Codes: G10L 19/005 Correction of errors induced by the transmission channel, if related to the coding algorithm
• CPC Class Codes: G10L 19/10 the excitation function being a multipulse excitation
• CPC Class Codes: H03M 13/09 Error detection only, e.g. using cyclic redundancy check [CRC] codes or single parity bit
• CPC Class Codes: H03M 13/23 using convolutional codes, e.g. unit memory codes
• CPC Class Codes: H03M 13/35 Unequal or adaptive error protection, e.g. by providing a different level of protection according to significance of source information or by adapting the coding according to the change of transmission channel characteristics
• CPC Class Codes: H04B 1/66 for reducing bandwidth of signals; for improving efficiency of transmission H04B1/68 takes precedence
• CPC Class Codes: H04L 1/0054 Maximum-likelihood or sequential decoding, e.g. Viterbi, Fano, ZJ algorithms
• CPC Class Codes: H04L 1/0057 Block codes H04L1/0061, H04L1/0064 take precedence
• CPC Class Codes: H04L 1/0065 Serial concatenated codes
• CPC Class Codes: H04L 1/20 using signal quality detector
• CPC Class Codes: H04L 1/201 Frame classification, e.g. bad, good or erased frame indication per se H04L1/0082
• CPC Class Codes: H04L 2001/0098 Unequal error protection

Abstract

A low-overhead method of protecting multi-pulse speech coders from the effects of severe random or fading pattern bit errors combines a standard error correcting code (convolutional rate 1/2 coding and Viterbi trellis decoding) for protection in random errors with cyclic redundancy code (CRC) error detection for fading errors. Compensation for detected fading errors takes place within the speech coder. Protection is applied only to the perceptually significant bits in the transmitted frame.

Claims

1. A method for protecting multi-pulse linear predictive speech coders from fading and random pattern bit errors comprising the steps of:

selecting perceptually significant bits of multi-pulse coded speech parameters to be protected, the parameters being a set of line spectrum pair frequency coefficient bits, pitch lag bits, voiced/unvoiced bits, pitch tap gain bits, pulse amplitude bits, and pulse position bits;

generating checksums of said selected bits;

combining said checksums and said selected bits;

coding the selected bits and said checksums according to a convolutional code to protect transmitted data from random errors;

combining the convolutionally coded selected bits, checksums, and nonselected bits of the multi-pulse encoded speech; and

transmitting the combined convolutionally coded selected bits, checksums, and nonselected bits.

2. The method recited in claim 1 further comprising the steps of:

receiving said combined convolutionally coded selected bits, checksums, and nonselected bits;

separating the combined convolutionally coded selected bits, checksums, and nonselected bits;

decoding said convolutionally coded selected bits;

merging said decoded selected bits with said nonselected bits to form a multi-pulse encoded speech bit stream; and

decoding said multi-pulse speech bit stream.

3. The method recited in claim 2 further comprising, prior to the step of coding the selected bits, the steps of:

generating checksums of said selected bits; and

combining said checksums and said selected

said method further comprising, prior to the step of merging said decoded selected bits with said nonselected bits, the steps of;

separating the decoded checksums and the decoded selected bits;

checking for errors based on said checksums; and

producing an indication of a detected error.

4. The method recited in claim 3 further comprising, when an error is detected in linear predictive coding/line spectrum pair frequencies or in pitch lag, the step of using all linear predictive coding coefficients contained in the most recent error-free received set of multi-pulse speech bits until a new error-free set of multi-pulse speech bits is received.

5. The method recited in claim 3 further comprising, when an error is detected in excitation data during the first two subframes of any frame, each subframe including a voiced/unvoiced bit, pitch tap gain bits, and amplitude and position bits for a plurality of pulses, the steps of:

using the most recently received error-free voiced/unvoiced bit for making a voiced/unvoiced decision;

zeroing all pulse amplitudes for which an error has been detected; and

setting a pitch tap gain for the multi-pulse speech bit stream to a predetermined value less than one.

6. The method recited in claim 3 further comprising, when an error is detected in excitation data during the last two subframes of any frame, each subframe including a voiced/unvoiced bit, pitch tap gain bits, and amplitude and position bits for a plurality of pulses, the steps of:

using the most recently received error-free voiced/unvoiced bit for making a voiced/unvoiced decision;

zeroing all pulse amplitudes for which an error has been detected; and

setting a pitch tap gain for the multi-pulse speech bit stream to a predetermined value less than one.

7. A multi-pulse speech coder/decoder system which minimizes the effects of fading and random pattern bit errors, said system comprising transmitter means including:

frame sorter means for selecting perceptually significant bits of multi-pulse encoded speech to be protected;

a convolutional encoder coupled to said frame sorter means for coding the selected bits to protect transmitted data from random errors; and

a multiplexer coupled to said frame sorter and said convolutional encoder for combining the convolutionally coded selected bits and nonselected bits of the multi-pulse encoded speech for transmission to receiver means.

8. The multi-pulse speech coder/decoder system recited in claim 7 wherein said transmitter means further includes:

a checksum generator coupled to said frame sorter means for generating checksums of said selected bits; and

a second multiplexer coupled to said frame sorter means and said checksum generator for combining said checksums and said selected bits and for supplying said combined checksums and said selected bits to said convolutional encoder.

9. The multi-pulse speech coder/decoder system recited in claim 7 and further comprising receiving means for receiving said combined convolutionally coded selected bits and nonselected bits, said receiving mans including:

a demultiplexer for separating the combined convolutionally coded selected bits and nonselected bits;

a decoder coupled to said demultiplexer for decoding said convolutionally coded selected bits;

frame merger means for merging said decoded selected bits with said nonselected bits to form a multi-pulse encoded speech bit stream; and

multi-pulse speech decoder means for decoding said multi-pulse speech bit steam, and for retaining the last received error-free multi-pulse encoded speech parameters and replacing a received signal with the multi-pulse encoded speech parameters upon receipt of a fading signal.

10. The multi-pulse speech coder/decoder system recited in claim 9 wherein said receiving means further includes:

a second demultiplexer coupled to said decoder for separating the decoded checksums and the decoded selected bits; and

a checksum generator coupled to said second demultiplexer for checking for errors based on said checksums and for supplying an indication of a detected error to said multi-pulse speech decoder means.

11. The multi-pulse speech coder/decoder system recited in claim 10 wherein said decoder comprises a Viterbi decoder.