Vector quantizer

Patent metadata

Field	Value
Publication number	US-9401155-B2
Application number	US-201214387716-A
Country	US
Kind code	B2
Filing date	Dec 12, 2012
Priority date	Mar 29, 2012
Publication date	Jul 26, 2016
Grant date	Jul 26, 2016

Abstract

Vector Quantizer and method therein for efficient vector quantization, e.g. in a transform audio codec. The method comprises comparing an input target vector s with a plurality of centroids, each centroid representing a respective class of codevectors in a codebook. Further, a starting point for a search related to the input target vector in the codebook is determined, based on the result of the comparison. The codevectors in the codebook are sorted according to a distortion measure reflecting the distance between each codevector and the centroids of the classes. The Vector Quantizer and method enables that the class of codevectors comprising the most probable candidate codevectors in regard of the input vector. s may be searched first.

First claim

The invention claimed is:

1. A method in a Vector Quantizer, the method comprising:
comparing an input target vector representing a spectral region of a segment of an audio signal with a plurality of centroids, each centroid representing a respective class of codevectors in a codebook,
determining a starting point for a search related to the input target vector in the codebook, based on the a result of the comparing the input target vector with the plurality of centroids,
performing a search in the codebook, starting at the determined starting point, and identifying a codevector to represent the input target vector, wherein the search is performed in one class of codevectors in the codebook that is selected based on the result of the comparing the input target vector with the plurality of centroids,
generating, using the codevector that is identified, a coded spectral region representation of the input target vector representing the spectral region of the segment of the audio signal, and
transmitting, through a communication circuit, the coded spectral region representation of the input target vector representing the spectral region of the segment of the audio signal,
wherein the codevectors in the codebook are sorted according to a distortion measure reflecting a distance between each codevector among the codevectors in the codebook and the centroids.

2. The method according to claim 1 , wherein a number of input target vectors per coding unit is variable.

3. The method according to claim 1 , wherein the search is performed within a search region determined based on a number of received input target vectors and a computational complexity constraint.

4. The method according to claim 1 , further comprising:
adapting the size of a search region in the codebook based on a number of input target vectors and a computational complexity constraint.

5. The method according to claim 3 , wherein the computational complexity constraint is set dynamically.

6. A transform codec comprising a Vector Quantizer, the Vector Quantizer comprising:
a comparing circuit, adapted to compare an input target vector representing a spectral region of a segment of an audio signal with a plurality of centroids, each centroid representing a respective class of codevectors in a codebook;
a determining circuit adapted to determine a starting point for a search in the codebook, based on a result from the compare of the input target vector with the plurality of centroids;
a search circuit, adapted to perform a search in the codebook, starting at the determined starting point, and identifying a codevector to represent the input target vector, wherein the search is performed in one class of codevectors in the codebook that is selected based on the result of the comparing the input target vector with the plurality of centroids; and
a encoder circuit, adapted to generate, using the codevector that is identified, a coded spectral region representation of the input target vector representing the spectral region of the segment of the audio signal,
a communication circuit, adapted to transmit the coded spectral region representation of the input target vector representing the spectral region of the segment of the audio signal,
wherein the codevectors in the codebook are sorted according to a distortion measure reflecting the distance between each codevector among the codevectors in the codebook and the centroids.

7. The transform codec according to claim 6 , further adapted to accept a variable number of input target vectors per coding unit.

8. The transform codec according to claim 6 , wherein the search is performed within a search region determined based on a number of input target vectors and a computational complexity constraint.

9. The transform codec according to claim 8 , further adapted to determine a search region in the codebook, based on a number of received input target vectors and a computational complexity constraint.

10. The transform codec according to claim 8 , wherein the computational complexity constraint is set dynamically.

11. A mobile terminal comprising the transform codec according to claim 6 .

12. A computer program comprising computer readable code on a non-transitory computer readable medium, which when run in a processor, causes a Vector Quantizer to perform the corresponding method according to claim 1 .

13. An audio codec device configured to encode an audio signal, the audio codec device comprising:
a processor; and
a memory coupled to the processor,
wherein the processor performs operations comprising:
comparing an input target vector representing a spectral region of a segment of the audio signal with a plurality of centroids, each centroid representing a respective class of codevectors in a codebook, wherein one or more codevectors in the codebook are stored in the memory;
determining a starting point for a search in the codebook related to the input target vector, based on the comparing the input target vector with the plurality of centroids;
performing the search in the codebook in the memory, starting at the determined starting point, and identifying a codevector to represent the input target vectors, wherein the search is performed in one class of codevectors in the codebook that is selected based on the result of the comparing the input target vector with the plurality of centroids; and
generating an coded spectral region representation of the input target vectors using the codevector that is identified,
transmitting, to an audio decoder, the coded spectral region representation of the input target vector representing the spectral region of the segment of the audio signal,
wherein the codevectors in the codebook in the memory are sorted according to a distortion measure reflecting a distance between each codevector among the codevectors in the codebook and the centroids.

14. The audio codec device of claim 13 ,
wherein the one or more codevectors in the codebook that are stored in the memory comprise a first portion of the codevectors in the codebook,
wherein a second portion of the codevectors in the codebook excluding the first portion, are not stored in the memory, and
wherein the second portion of the codevectors are a flipped version of the first portion of the codevectors.

15. The audio codec device of claim 14 ,
wherein the search in the codebook is performed on the first portion of the codevectors in the memory.

16. The audio codec device of claim 14 , further comprising:
generating the second portion of the codevectors in the codebook based on the first portion,
wherein the performing the search in the codebook comprises performing the search on the second portion of the codevectors in the codebook that were generated.

17. The method according to claim 1 ,
wherein the starting point is one of a codevector that is a maximum distance from one of the plurality of centroids or a codevector that is a minimum distance from one of the plurality of centroids.

18. The method according to claim 1 , wherein the performing the search comprises:
first performing the search in the one class of codevectors in the codebook that is selected; and
subsequently performing the search in other classes of codevectors in the codebook not including the one class of codevectors that were first searched.

19. The method according to claim 1 , wherein the search space is dynamically adjusted by increasing or reducing a size of the search space based on the number of target vectors.

20. The method a

Assignees

• Ericsson Telefon Ab L M
• ERICSSON TELEFON AB L M (publ)

Inventors

• Grancharov Volodya
• Jansson Toftgård Tomas

Classifications

• H04N19/94

  Vector quantisation · CPC title

• H03M7/3082

  Vector coding (for television signals, see H04N19/94) · CPC title

• G10L19/038 Primary

  Vector quantisation, e.g. TwinVQ audio · CPC title

• G10L2019/0013

  Codebook search algorithms · CPC title

• G10L19/18

  Vocoders using multiple modes · CPC title