Method and system for the arrangement of vocoder data for the masking of transmission channel induced errors

• US 5,600,754 A
• Filed: 02/14/1994
• Issued: 02/04/1997
• Est. Priority Date: 01/28/1992
• Status: Expired due to Term

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Gardner, William R., Jacobs, Paul E.
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): Chawan, Vijay B.
• Application Number:
  US08/194,824
• Time in Patent Office: 1,086 Days
• Field of Search: 395/2.3, 395/2.31, 395/2.33
• US Class Current: 704/221
• CPC Class Codes: G10L 19/005 Correction of errors induced by the transmission channel, if related to the coding algorithm
• CPC Class Codes: G10L 19/07 Line spectrum pair [LSP] vocoders
• 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 1/665 using psychoacoustic properties of the ear, e.g. masking effect
• CPC Class Codes: H04L 1/0071 Use of interleaving interleaving per se H03M13/27
• CPC Class Codes: H04L 1/0083 Formatting with frames or packets; Protocol or part of protocol for error control
• CPC Class Codes: H04L 1/0086 Unequal error protection H04L27/00 and H04L1/004 take precedence for layer 1/2 aspects, e.g. bit loading

Abstract

A method and system by which parameter data representative of vocoded speech are organized into a data packet for transmission so as to reduce the impact of transmission channel induced errors on the data packet. A data packet is constructed with certain most perceptually significant bits of parameter data at the beginning of the data packet. Following in the data packet are lesser perceptually significant bits of the same parameter data. Other parameter data then follows in the data packet. Interleaved in the data packet following the most perceptually significant bits at the beginning of the data packet are most perceptually significant bits of other parameter data. A parity check code is computed from the most perceptually significant bits in the data packet and also interleaved in the data packet following the most perceptually significant bits at the beginning of the data packet.

Claims

1. In a vocoding system a method for formatting vocoded speech parameter data corresponding to a frame of speech samples into a data packet for transmission comprising the steps of:

receiving vocoded speech parameter data wherein certain data of said vocoded speech parameter data is of greater relative importance than other data of said vocoded speech parameter data;

providing a certain portion of said certain data at a beginning of a data packet;

providing, exclusively within a remaining packet portion of said data packet following said certain portion of said certain data, another certain portion of said certain data; and

providing, exclusively within said remaining packet portion, a remaining portion of said certain data interleaved within said other data.

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

computing an error correcting code from said certain data of said vocoded speech parameter data; and

interleaving said error correcting code within said interleaved remaining portion of said certain data and other data within said data packet.

3. The method of claim 2 wherein said another portion of said certain data comprises data of a second parameter type and data of a third parameter type and wherein said other data comprises data of a third data type and data of a fourth parameter type and wherein said step of providing said remaining portion of said certain data following said certain portions of one of said certain speech parameter data set in said data packet comprises the steps of:

providing said data of said third parameter type immediately following said another portion of one of said certain data set in said data packet; and

providing said data of said fourth parameter type immediately following said data of said third parameter type in said data packet.

4. The method of claim 1 wherein said certain portion is of a first parameter type and wherein said another certain portion comprises data of said first parameter type and data of a second parameter type and wherein said step of providing said another certain portion of said certain data comprises the steps of:

providing said data of said first parameter type immediately following said certain portions of certain data in said data packet; and

providing said data of said first parameter type immediately following said data of said first parameter type in said data packet.

5. In a vocoding system in which a variable rate linear prediction coding (LPC) vocoder encodes frames of speech samples to produce corresponding sets of encoded speech parameter data based upon an analysis of speech samples in each speech sample frame and predetermined subframes thereof, a method for formatting a frame of one rate parameter data for transmission to reduce the impact of transmission channel induced errors in a transmitted frame of parameter data as received and reconstructed into a corresponding frame of speech samples, said method comprising the steps of:

receiving Line Spectral Pair (LSP) frequency data;

receiving pitch data;

receiving codebook data;

computing a parity check code from certain ones of said LSP frequency data, said pitch data and said codebook data;

assembling in a predetermined sequence within a beginning packet portion of a data packet a first portion of each one of said LSP frequency data;

assembling in a predetermined sequence, other than within beginning packet portion, a second portion of each one of said LSP frequency data, said sequence of second portion of LSP frequency data following said sequence of first portion of said LSP frequency data in said data packet;

providing, other than within said beginning packet portion, a sequence of said pitch data and said codebook data following said sequence of second portion of LSP frequency data in said data packet; and

providing said parity check code within said data packet other than within said beginning packet portion.

6. The method of claim 5 further comprising the step of interleaving said parity check code within said data packet following said first portion of each one of said LSP frequency data.

7. The method of claim 6 further comprising the step of interleaving within said data packet a portion of each codebook data within said sequence of second portion of each one of said LSP frequency data and within said sequence of said pitch data and said codebook data, and wherein said parity check codes are interleaved within said sequence of pitch data and codebook data following said interleaved portions of codebook data.

8. The method of claim 5 wherein at said one rate said first plurality of sets of pitch data comprises four sets of pitch data with each set of pitch data comprising pitch lag and pitch gain data, and said first plurality of sets of codebook data comprises eight sets of codebook data with each set of codebook data comprising codebook index and codebook gain data.

9. The method of claim 5 for formatting a frame of another rate parameter data for transmission further comprising the steps:

receiving second Line Spectral Pair (LSP) frequency data;

receiving second pitch data;

receiving second codebook data;

assembling in a predetermined sequence at a beginning of a data packet each one of said LSP data of said second LSP data; and

assembling in a predetermined sequence said second pitch data and codebook data following said sequence of LSP data in said data packet.

10. In a vocoding system in which a vocoder produces variable rate frames of parameter data and corresponding frame rate indication, a data packing system for assembling each frame of parameter data into a data packet comprising:

parity check bit computation means for receiving a frame rate indication and predetermined portions of a frame of parameter data, said parity check bit computation means responsive to a highest frame rate indication for computing and providing an output of parity check bit data; and

data packing means for receiving said frame rate indication, said frame of parameter data and said parity check bit data said data packing means responsive to said frame rate indication for organizing said frame of parameter data for output according to a predetermined format in a data packet, wherein at said highest frame rate indication said data packing means arranges said data packet such that a sequence of a portion of each parameter data of one parameter data type is provided in a beginning packet portion, and a sequence of a remaining portion of each parameter data of said one parameter data type is provided, exclusively within a remaining packet portion following said beginning packet portion, with remaining parameter data of said frame of parameter data and said parity check bit data.

11. The system of claim 10 wherein said data packing means in response to said highest frame rate indication is further for arranging said data packet with a portion of said remaining parameter data and said parity check bit data interleaved throughout said remaining portion of each parameter data of said one parameter data type and said remaining parameter data.

12. The data packing system of claim 10 wherein said data packing means is responsive to rate indications other than said highest frame rate indication for arranging said data packet beginning with a data sequence of one parameter data type followed by data sequences of remaining parameter data types.

13. The system of claim 10 wherein said data packing means comprises:

frame buffer means for receiving and storing said frame of parameter data, for receiving a first output signal and providing an output of selected parameter data and portions thereof;

packet buffer means for receiving and storing said output of parameter data and portions thereof from said frame buffer and said parity check bit data from said parity check bit computation means, and for providing an output of said selected parameter data and portions thereof as said data packet; and

packing means for receiving said rate indication and generating a said first output Signal for controlling the output of data from said first buffer means.

14. In a vocoding system in which received variable data rate frames of vocoded speech parameter data are decoded into corresponding frames of speech samples, a method for reordering said parameter data as formatted in a data packet within each received variable data rate frame wherein parameter data of a highest rate frame contains a portion of each parameter data of one parameter data type in a predetermined order within a beginning packet portion of the data packet, said beginning packet portion being followed by a remaining packet portion within which is exclusively provided another portion of each parameter data of the one set of parameter data in a predetermined order, and followed exclusively within said remaining packet portion by remaining parameter data of said variable data frame, and having parity data within said exclusively within said remaining packet portion, said parity data being computed from parameter data within said data packet, said method comprising the steps of:

receiving a highest rate frame of parameter data;

reassembling corresponding ones of said portion and said another portion of each parameter data of said one parameter data type to provide reassembled data;

computing parity data from parameter data in said received highest rate frame of parameter data;

comparing said computed parity data with said parity data in said received highest rate frame of parameter data; and

providing said reassembled data and said remaining parameter data for the reproduction of said speech samples if said computed parity data is the same as said received parity data.

15. The method of claim 14, wherein lower rate frames of parameter data are formatted with parameter data in a data packet in a predetermined order corresponding to the data rate of said variable data rate frame, further comprising the steps of:

receiving an indication of a data rate for each received variable data rate frame of parameter data; and

providing parameter data in a predetermined format according to said received variable data rate frame data rate indication.

16. The method of claim 14, wherein said highest rate frame of parameter data further has a portion of each parameter data of another parameter data type interleaved within said another portion of parameter data of said one parameter data type, further comprising the step of reassembling each portion of parameter data of said another parameter data type with a corresponding remaining portion of each of said parameter data of said another parameter data type.

17. The method of claim 14 further comprising the steps of:

receiving a full rate likely indication for said received highest rate frame of parameter data, said full rate likely indicative of at least one error in one of said parameter data and said parity data;

computing parity data from parameter data in said received highest rate frame of parameter data;

comparing said computed parity data with said parity data in said received highest rate frame of parameter data;

detecting from said comparison of said computed parity data with said parity data in said received highest rate frame of parameter data an error in at least one of said parameter data and said parity data in said received highest rate frame of parameter data;

correcting said detected error if occurring in certain ones of said parameter data and said parity data in said received highest rate frame of parameter data;

reassembling corresponding ones of said portions and said another portions of each parameter data of said one set of parameter data; and

providing an output of said parameter data when said detected error occurs in said certain ones of said parity data and from corrected parameter data when said detected error occurs in said certain ones of said parameter data.

18. In a vocoding system in which received variable rate frames of data contain vocoded speech parameter data, with the frame data formatted according to a predetermined format, and the parameter data is decoded into corresponding frames of speech samples, a system for reorganizing parameter data within each received variable rate frame of parameter data comprising:

input means for receiving and storing a frame of data, and for providing an ordered sequence of data of said received frame of data in accordance with a control signal; and

unpacking means for providing said control signal in accordance with a frame rate indication for said received frame of data.

19. The system of claim 18 wherein a highest rate frame of data contains a portion of each parameter data of one parameter data type in a predetermined order at a beginning of the data packet followed by another portion of each parameter data of said one parameter data type in a predetermined order, followed by remaining parameter data, and having parity data within said data packet computed from parameter data within said data packet, said system further comprising:

parity check means for receiving said parity data in a received highest rate frame of parameter data, computing parity data from parameter data in said received highest rate frame, comparing said computed parity data with said received highest rate frame parity data, and providing an indication when said computed parity data is the same as said received parity data; and

wherein said input means is responsive to said control signal and said indication for providing an output of said parameter data comprised of reassembled corresponding ones of said portions and said another portions of each parameter data of said of parameter data type.

20. The system of claim 19 wherein said highest rate frame of parameter data further has a portion of each parameter data of another parameter data type interleaved within said another portion of parameter data of said one parameter data type, said input means further responsive to said control signal for providing an output of each portion of parameter data of said another parameter data type with a corresponding remaining portion of each of said parameter data of said another parameter data type.

21. The system of claim 19 wherein:

said unpacking means and said parity check means further receive a full rate likely indication for said received highest rate frame of parameter data, said full rate likely indicative of at least one error in at least one of said parameter data and said parity data;

said parity check means further for detecting from said comparison of said computed parity data with said parity data in said received highest rate frame of parameter data an error in at least one of said parameter data and said parity data in said received highest rate frame of parameter data, correcting said detected error if occurring in certain ones of said parameter data and said parity data in said received highest rate frame of parameter data; and

wherein said input means is responsive to said control signal and said indication for providing an output of said parameter data when said detected error occurs in said certain ones of said parity data, and an output of corrected parameter data when said detected error occurs in said certain ones of said parameter data, and where said output parameter data and corrected parameter data is comprised of reassembled corresponding ones of said portions and said another portions of each parameter data of said one set of parameter data.

22. A method for providing a frame of full rate variable rate vocoder data from received speech samples, said method comprising the following steps of:

generating speech parameter data representative of the received speech samples;

providing, from the speech parameter data, an ordered sequence of most significant line spectral pair (LSP) bits;

providing, from the speech parameter data, an interleaved sequence of less significant LSP bits and certain ones of most significant codebook gain (CBGAIN) bits; and

providing, from the speech parameter data, an interleaved sequence of remaining CBGAIN bits, pitch gain (PGAIN) bits, pitch lag (PLAG) bits, codebook index (CBINDEX) bits and parity check bits (PCB).

23. A method for providing a frame of half rate variable rate vocoder data from received speech samples, said method comprising the following ordered steps of:

generating speech parameter data representative of the received speech samples;

providing, from the speech parameter data, an ordered sequence of line spectral pair (LSP) bits; and

providing, from the speech parameter data, an ordered sequence of codebook gain (CBGAIN) bits, pitch gain (PGAIN) bits, pitch lag (PLAG) bits and codebook index (CBINDEX).

24. A method for providing a frame of quarter rate variable rate vocoder data from received speech samples, said method comprising the following ordered steps of:

generating speech parameter data representative of the received speech samples;

providing, from the speech parameter data, an ordered sequence of line spectral pair (LSP) bits; and

providing, from the speech parameter data, an ordered sequence of codebook gain (CBGAIN) bits, pitch gain (PGAIN) bits, pitch lag (PLAG) bits and codebook index (CBINDEX).

25. A method for providing a frame of eighth rate variable rate vocoder data from received speech samples, said method comprising the following ordered steps of:

generating speech parameter data representative of the received speech samples;

providing, from the speech parameter data, an interleaved sequence of codebook seed (CBSEED) bits and line spectral pair (LSP) bits; and

providing, from the speech parameter data, an ordered sequence of codebook gain (CBGAIN) bits.

26. A vocoder comprising:

vocoder engine means for receiving a frame of speech samples and providing speech parameter data representing said frame of speech samples in accordance with a predetermined speech model; and

ordering means for receiving said speech parameter data and providing a packet of speech parameter data wherein most perceptually significant portions of said speech parameters are provided in a first part of said packet followed by a second part of said packet within which remaining portions of said speech parameter data are exclusively provided, wherein portions of said remaining portions of speech parameter data are interleaved exclusively throughout said second part of said packet in accordance with at least one predetermined format.

27. The vocoder of claim 26 wherein said vocoder engine means comprises:

means for providing linear predictive code (LPC) data;

means for providing pitch data; and

means for providing codebook data.

28. The vocoder of claim 27 further comprising means for converting said LPC data to line spectral pair (LSP) data.

29. The vocoder of claim 28 wherein said ordering means provides said LSP data first in said packet.

30. The vocoder of claim 26 wherein said vocoder engine means provides said parameter data at a selected data rate and provides a data rate signal indicative of said selected data rate and wherein said ordering means is responsive to said data rate signal.

31. In a communication system wherein data from a frame of data is reordered to provide more significant portions of said data in a first half of a transmission packet and wherein remaining portions of said data are provided exclusively in a second half of said transmission packet distinct from said first half of said transmission packet, and wherein portions of said remaining data are interleaved exclusively throughout said second half of said transmission packet, a subsystem for reordering said frame of data comprising:

control means for providing a control signal based upon a frame rate associated with said transmission packet, said control signal being indicative of an unpacking order for said transmission packet; and

input means for receiving said transmission packet and providing reordered data from said transmission packet in accordance with said control signal.

32. The subsystem of claim 31 wherein said transmission packet further comprises at least one parity bit, said subsystem further comprising a parity check means for receiving said transmission packet and determining in accordance with a predetermined parity format at least one check bit, comparing said at least one check bit with said at least one parity bit and providing a signal indicative of the equality of said at least one parity bit and said at least one check bit.

33. In a communication system wherein a frame of speech samples is encoded according to a speech model to provide a frame of parameter data and wherein said frame of parameter data is reordered to provide more significant portions of said parameter data in a first half of a transmission packet and wherein remaining portions of said parameter data are provided exclusively in a second half of said transmission packet distinct from said first half of said transmission packet and wherein portions of said remaining data are interleaved exclusively throughout said second half of said transmission packet, a system for providing an estimate of said speech data from said transmission packet comprising:

ordering means for receiving said transmission packet and providing reconstructed estimates of said parameter data from said transmission packet, said reconstructed estimates of said parameter data being ordered in response to a frame rate associated with said transmission packet; and

vocoder means for receiving said reconstructed estimates of said parameter data and decoding said reconstructed estimates of said parameter data in accordance with said speech model to provide an estimate of said frame of speech samples.

34. The system of claim 33 wherein said transmission packet comprises parameter data at a selected data rate wherein said selected data rate is selected from a plurality of data rates and wherein said vocoder means is further for decoding said parameter data in accordance with said data rate.

35. In a vocoding system, a method for formatting speech parameter data into a data packet for transmission, said method comprising the steps of:

generating said speech parameter data in response to an input frame of speech samples wherein certain data of said speech parameter data is of greater relative importance than other data of said speech parameter data;

providing a certain portion of said certain data within a beginning packet portion of a data packet, said beginning packet portion of said data packet including exclusively said certain portion of said certain data;

providing interleaved data exclusively in a remaining packet portion of said data packet following said beginning packet portion, said interleaved data being formed by interleaving a remaining portion of said certain data with said other data.

36. The method of claim 35 further comprising the steps of:

computing an error correcting code from said certain data; and

interleaving said error correcting code exclusively within said interleaved data.