Apparatus for encoding digital data with reduction of perceptible noise

• US 5,502,789 A
• Filed: 03/12/1993
• Issued: 03/26/1996
• Est. Priority Date: 03/07/1990
• Status: Expired due to Term

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Akagiri, Kenzo
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): Doerrler, Michelle
• Application Number:
  US08/029,988
• Time in Patent Office: 1,110 Days
• Field of Search: 381/29-40, 395/2.1-2.39
• US Class Current: 704/204
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect

Abstract

Apparatus for efficiently encoding a digital signal divides the digital signal into a number of signals within respective predetermined frequency bands which are segmented into blocks which, in turn, are orthogonally transformed so as to form coefficient data. The coefficient data is then quantized. A detection circuit detects whether at least the divided signal within the lowest frequency band contains a transient. If a transient is detected, a control circuit controls the encoding processing of a least the divided signal within the lowest frequency band by, for example, increasing the number of bits used for quantizing, reducing the block size, and/or reducing the signal level so as to reduce the perceptibility of noise in the processed signal.

Claims

1. Apparatus for encoding a digital signal comprised of a plurality of samples, said apparatus comprising:

frequency dividing means for dividing said digital signal comprised of said plurality of samples into a plurality of frequency divided signals within respective frequency bands;

encoding means for encoding said frequency divided signals including block segmentation means for segmenting each of said frequency divided signals into respective size blocks each having a respective number of said samples, orthogonal transformation means for orthogonally transforming said blocks to form coefficient data, and quantization means for quantizing said coefficient data with a number of bits to form an encoded digital signal;

transient detection means receiving at least the one of said plurality of frequency divided signals within the lowest of said frequency bands for detecting a transient in the received frequency divided signal within the lowest frequency band and for generating an output signal indicating that said transient has been detected; and

control means responsive to said output signal from said transient detection means for controlling the encoding performed by said encoding means of at least said frequency divided signal within said lowest frequency band so as to reduce the perceptibility of noise in the encoded digital signal.

2. Apparatus according to claim 1, wherein said control means includes quantizing bit number control means for controlling the number of bits used by said quantization means for quantizing said coefficient data such that the number of bits used for quantizing at least said coefficient data formed from said frequency divided signal within the lowest frequency band is increased when said output signal from said transient detection means indicates that said transient has been detected.

3. Apparatus according to claim 2, wherein said control means further includes:

first memory means for storing data calculated by said quantization means in accordance with a spectral examination of said coefficient data which is representative of the number of bits to be used by said quantization means;

second memory means for storing data which differs from said data stored in said first memory means and which is representative of the number of bits to be used for quantizing at least said coefficient data formed from said frequency divided signal within said lowest frequency band when said output signal generated by said transient detection means indicates that said transient has been detected; and

switch means for connecting said second memory means to said quantization means in response to said output signal generated by said transient detection means and in the absence of said output signal for connecting said first memory means to said quantization means.

4. Apparatus according to claim 2, wherein said control means further includes:

memory means for storing, at respective addresses therein, a plurality of different bit number data calculated by said quantization means in accordance with a spectral examination of said coefficient data which respectively correspond to different degrees of transient change; and

address generator means receiving said output signal of said transient detection means and, in response thereto, generating address signals for supply to said memory means so as to cause said memory means to output the bit number data corresponding to the degree of transient change indicated by said output signal of said transient detection means for use by said quantization means.

5. Apparatus according to claim 1, wherein said control means includes block size control means for controlling the size of the blocks into which at least said frequency divided signal within the lowest frequency band is segmented such that the block size of at least said frequency divided signal within the lowest frequency band is decreased when said output signal from said transient detection means indicates that said transient has been detected.

6. Apparatus according to claim 1, in which each of said frequency divided signals within the respective frequency bands has a respective signal level and in which said control means includes signal level control means for controlling the level of at least said frequency divided signal within the lowest frequency band such that the level of at least said frequency divided signal within the lowest frequency band is decreased when said output signal from said transient detection means indicates that said transient has been detected.

7. Apparatus according to claim 1, wherein said transient detection means includes:

block forming means for segmenting the plurality of samples of said frequency divided signal within the lowest frequency band which are to be formed into one of said blocks of said signal within the lowest frequency band by said block segmentation means into a plurality of smaller size blocks;

energy calculating means for calculating respective energy values in said smaller size blocks and for generating output signals respectively corresponding to said energy values; and

processing means for processing said output signals of said energy calculating means so as to form said output signal of said transient detection means.

8. Apparatus according to claim 7, wherein said processing means includes a plurality of divider means for dividing predetermined pairs of said output signals of said energy calculating means so as to produce respective divided output signals.

9. Apparatus according to claim 8, further comprising synthesizing means for combining said divided output signals from said divider means to form said output signal of said transient detection means.

10. Apparatus according to claim 8, wherein said processing means further includes:

a plurality of comparator means each receiving a respective one of said divided output signals and each operative to compare the received one of said divided output signals with a reference input signal and to generate a predetermined output signal if said received one of said divided output signals is substantially equal to a predetermined multiple of said reference input signal;

logical addition gate means receiving said predetermined output signal from each of said comparator means for generating a logical output signal therefrom; and

latch means receiving said logical output signal of said logical addition gate means for producing said output signal of said transient detection means in binary form.

11. Apparatus according to claim 1, wherein said frequency dividing means divides said digital signal into said plurality of frequency divided signals within the respective frequency bands such that the one of said plurality of frequency divided signals within the highest frequency band has a bandwidth which is greater than the bandwidths of the other frequency divided signals.

12. Apparatus according to claim 11, in which said digital signal is an audio digital signal and in which said frequency dividing means divides said audio digital signal into signals within three frequency bands of substantially 0 to 6 kHz, 6 kHz to 12 kHz, and 12 kHz to 24 kHz bands, respectively.

13. Apparatus according to claim 1, wherein said frequency dividing means includes a plurality of quadrature mirror filters connected in cascade in which each of said quadrature mirror filters receives a respective input digital signal and divides said respective input digital signal into a pair of output digital signals each having a frequency range of half that of the respective input digital signal.

14. Apparatus according to claim 1, wherein said block segmentation means and said orthogonal transformation means include an orthogonal transformation circuit for each of said frequency bands.

15. Apparatus according to claim 1, wherein said orthogonal transformation means orthogonally transforms said blocks in accordance with fast Fourier transformations.

16. Apparatus for encoding a digital signal comprised of a plurality of samples, said apparatus comprising:

encoding means for encoding said digital signal including block segmentation means for segmenting said signal into blocks each having a respective number of said samples, transformation means for transforming said blocks from a time domain to a frequency domain so as to form coefficient data, and quantization means for quantizing said coefficient data with a number of bits to form an encoded signal;

transient detection means for detecting a transient in the received digital signal, including means receiving said blocks of samples for dividing said blocks into subblocks, means for determining a signal characteristic for each subblock, means for comparing the respective determined signal characteristics of the subblocks making up one of said blocks and means for generating an output signal indicating that said transient has been detected on the basis of a comparison of the respective determined signal characteristics; and

control means responsive to said output signal from said transient detection means for controlling the encoding of said digital signal performed by said encoding means so as to reduce the perceptibility of noise in the encoded signal.

17. Apparatus according to claim 16, wherein said control means includes block size control means for controlling the size of the blocks into which said digital signal is segmented such that the number of said samples in the block is decreased when the output signal from said transient detection means indicates that said transient has been detected.

18. Apparatus for encoding a digital signal comprised of a plurality of samples, said apparatus comprising:

frequency dividing means for dividing said digital signal comprised of said plurality of samples into a plurality of frequency divided signals within respective frequency bands; and

encoding means for encoding said frequency divided signals including block segmentation means for segmenting each of said frequency divided signals into respective size blocks each having a respective number of samples, transformation means for transforming said blocks from a time domain to a frequency domain so as to form coefficient data, and quantization means for quantizing said coefficient data with a number of bits to form an encoded digital signal.

19. Apparatus according to claim 18, wherein said block segmentation means segments the frequency divided signal within at least one of said frequency bands into blocks which each have a respective number of samples which is larger than a respective number of samples of the blocks into which said block segmentation means segments the frequency divided signals that are not within said at least one of said frequency bands.

20. Apparatus according to claim 19, wherein said block segmentation means segments the frequency divided signal within said at least one of said frequency bands into blocks which have twice as many samples as said blocks into which said block segmentation means segments the frequency divided signals that are not within said at least one of said frequency bands.

21. Apparatus according to claim 20, wherein said block segmentation means segments the frequency divided signal within said at least one of said frequency bands into blocks which consist of 128 samples and said block segmentation means segments the frequency divided signals that are not within said at least one of said frequency bands into blocks which consist of 64 samples.

22. Apparatus according to claim 18, wherein said block segmentation means segments the frequency divided signal within a at least one of said frequency bands into blocks which each have a respective time duration that is longer than respective time durations of the blocks into which said block segmentation means segments the frequency divided signals that are not within said at least one of said frequency bands.

23. Apparatus according to claim 22, wherein said block segmentation means segments the frequency divided signal within said at least one of said frequency bands into blocks which each have a respective time duration that is at least twice as long as respective time durations of the blocks into which said block segmentation means segments the frequency divided signals that are not within said at least one of said frequency bands.

24. Apparatus according to claim 18, wherein said frequency dividing means divides said digital signal into said plurality of frequency divided signals within the respective frequency bands such that the one of said plurality of frequency divided signals within the highest frequency band has a bandwidth which is greater than the bandwidths of the other frequency divided signals.

25. Apparatus according to claim 18, wherein said digital signal is an audio digital signal and said frequency dividing means divides said audio digital signal into signals within three frequency bands consisting of a low band, a middle band and a high band.

26. Apparatus according to claim 25, wherein said low band is substantially 0 to 6 kHz, said middle band is substantially 6 kHz to 12 kHz, and said high band is substantially 12 kHz to 24 kHz.

27. Apparatus according to claim 25, wherein said block segmentation means segments said signal within said low band into blocks which each consist of 128 samples and said block segmentation means segments said signals within said middle and high bands, respectively, into blocks which each consist of 64 samples.

28. Apparatus according to claim 25, wherein said block segmentation means segments said signal within at least one of said bands into blocks which each have a duration of approximately 10.67 ms, and said block segmentation means segments said signal within another of said bands into blocks which each have a duration of approximately 2.67 ms.

29. Apparatus according to claim 25, wherein said block segmentation means segments said signal within one of said bands into blocks which each have a duration which is twice as long as a duration of the blocks into which said block segmentation means segments said signal within the other bands.

30. Apparatus according to claim 18, wherein said frequency dividing means includes a plurality of quadrature mirror filters connected in cascade in which each of said quadrature mirror filters receives a respective input digital signal and divides said respective input digital signal into a pair of output digital signals each having a frequency range of half that of the respective input digital signal.

31. Apparatus according to claim 30, wherein said plurality of quadrature mirror filters consists of a first quadrature mirror filter and a second quadrature mirror filter, said first quadrature mirror filter receiving said digital signal and dividing said digital signal into a high band signal and a base band signal, and said second quadrature mirror filter receiving said base band signal and dividing said base band signal into an intermediate band signal and a low band signal.

32. Apparatus according to claim 31, wherein said high band signal has a bandwidth of approximately 12 kHz, and said intermediate and low band signals each have a bandwidth of approximately 6 kHz.

33. Apparatus according to claim 31, wherein said high band has a bandwidth which is twice as wide as a bandwidth of said low band.

34. A method of encoding a digital signal comprised of a plurality of samples, comprising the steps of:

dividing said digital signal comprised of said plurality of samples into a plurality of frequency divided signals within respective frequency bands;

segmenting each of said frequency divided signals into respective size blocks each having a respective number of samples,

transforming said blocks from a time domain to a frequency domain so as to form coefficient data, and

quantizing said coefficient data with a number of bits to form an encoded signal.

35. A method according to claim 34, wherein the frequency divided signal within at least one of said frequency bands is segmented into blocks which each have a respective number of samples which is larger than a respective number of samples of the blocks resulting from segmentation of the frequency divided signals that are not within said at least one of said frequency bands.

36. A method according to claim 35, wherein the frequency divided signal within said at least one of said frequency bands is segmented into blocks which have twice as many samples as the blocks resulting from segmentation of the frequency divided signals that are not within said at least one of said frequency bands.

37. A method according to claim 36, wherein the frequency divided signal within said at least one of said frequency bands is segmented into blocks which consist of 128 samples and the blocks resulting from segmentation of the frequency divided signals that are not within said at least one of said frequency bands consist of 64 samples.

38. A method according to claim 34, wherein the frequency divided signal within at least one of said frequency bands is segmented into blocks which each have a respective time duration that is longer than respective time durations of the blocks resulting from segmentation of the frequency divided signals that are not within said at least one of said frequency bands.

39. A method according to claim 38, wherein the frequency divided signal within said at least one of said frequency bands is segmented into blocks which each have a respective time duration that is at least twice as long as respective time durations of the blocks resulting from segmentation of the frequency divided signals that are not within said at least one of said frequency bands.

40. A method according to claim 34, wherein said digital signal is divided into said plurality of frequency divided signals within the respective frequency bands such that the one of said plurality of frequency divided signals within the highest frequency band has a bandwidth which is greater than the bandwidths of the other frequency divided signals.

41. A method according to claim 34, wherein said digital signal is an audio digital signal and is divided during said dividing step into signals within three frequency bands consisting of a low band, a middle band and a high band.

42. A method according to claim 41, wherein said low band is substantially 0 to 6 kHz, said middle band is substantially 6 kHz to 12 kHz, and said high band is substantially 12 kHz to 24 kHz.

43. A method according to claim 41, wherein said segmenting step includes segmenting said signal within said low band into blocks which each consist of 128 samples and segmenting said signals within said middle and high bands, respectively, into blocks which each consist of 64 samples.

44. A method of according to claim 41, wherein said segmenting step includes segmenting said signal within at least one of said bands into blocks which each have a duration of approximately 10.67 ms, and segmenting said signal within another of said bands into blocks which each have a duration of approximately 2.67 ms.

45. A method according to claim 41, wherein said segmenting step includes segmenting said signal within one of said bands into blocks which each have a duration which is twice as long as a duration of the blocks resulting from segmentation of said signal within said other bands.