Apparatus, method, and system for compressing a digital input signal in more than one compression mode

• US 5,530,750 A
• Filed: 02/18/1994
• Issued: 06/25/1996
• Est. Priority Date: 01/29/1993
• 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): Cain, David C.
• Application Number:
  US08/198,863
• Time in Patent Office: 858 Days
• Field of Search: 380/3, 380/4, 380/28, 380/46, 380/49
• US Class Current: 704/500
• CPC Class Codes: G11B 20/00007 Time or data compression or expansion audio compression based on psychoacoustics G10L19/00; data processing for reproducing audio data at different playback speeds G10L21/04; video compression H04N19/00; data compression per se H03M7/30
• CPC Class Codes: G11B 20/10527 Audio or video recording; Data buffering arrangements G11B20/12 - G11B20/18 take precedence
• CPC Class Codes: G11B 27/031 Electronic editing of digitised analogue information signals, e.g. audio or video signals
• CPC Class Codes: G11B 27/3063 Subcodes
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect

Abstract

An apparatus, method, and system in which a compressed digital signal is derived from a digital input signal by compressing the digital input signal in a selected one of plural compression modes. The digital output signal has a different bit rate in each compression mode, but the digital input signal is received at the same bit rate in all the compression modes. In the method, spectral coefficients are derived from the digital input signal, and are grouped by frequency and by time into bands. Each band has a frequency width, and the frequency width of at least one of the bands is set according to the selected compression mode. The spectral coefficients in each band are quantized to provide quantized spectral coefficients. Finally, the quantized spectral coefficients in each band are included in the compressed digital signal, together with sub information for each band.

Claims

1. Apparatus for compressing a digital input signal in a selected one of plural compression modes to provide a compressed digital signal, the plural compression modes including a first compression mode and a second compression mode, the compressed digital signal having a higher bit rate in the first compression mode than in the second compression mode, the apparatus comprising:

deriving means for deriving from the digital input signal spectral coefficients grouped by frequency and time into bands, each band having a frequency width, the deriving means including frequency width setting means for setting the frequency width of at least one of the plural bands according to the selected compression mode;

a quantizer, the quantizer receiving the bands of spectral coefficients from the deriving means, and providing bands of quantized spectral coefficients; and

including means for including in the compressed digital signal the quantized spectral coefficients from the quantizer and sub information for each band.

2. The apparatus of claim 1, wherein, in the first compression mode, the frequency width setting means sets the frequency width of the at least one band to a first frequency width, and, in the second compression mode, sets the frequency width of the at least one band to a second frequency width, the second frequency width being greater than the first frequency width.

3. The apparatus of claim 2, wherein the second frequency width is an integral number of times greater than the first frequency width.

4. The apparatus of claims 2 or 3, wherein:

the digital input signal is a digital audio signal; and

the deriving means groups selected ones of the spectral coefficients by frequency into bands for determining quantizing noise and masking, a majority of the bands having frequency widths set to be broader towards higher frequencies.

5. The apparatus of claim 1, wherein:

the digital input signal is a digital audio signal; and

the deriving means groups selected ones of the spectral coefficients by frequency into bands for determining quantizing noise and masking, a majority of the bands having frequency widths set to be broader towards higher frequencies.

6. The apparatus of claim 5, wherein the deriving means includes means for receiving the digital input signal at a fixed sampling frequency, the sampling frequency being invariable between the compression modes.

7. The apparatus of claims 1 or 5, wherein the deriving means includes:

a low-pass filter having a cut-off frequency set according to the selected one of the plural compression modes, the low-pass filter receiving the digital input signal, and providing a frequency-limited signal, the cut-off frequency of the low-pass filter being set to a higher frequency in the first compression mode than in the second compression mode; and

means for deriving the spectral coefficients from the frequency-limited signal.

8. The apparatus of claim 7, wherein the deriving means additionally includes means for receiving the digital input signal a fixed sampling frequency, the sampling frequency being invariable between the compression modes.

9. The apparatus of claim 7, wherein:

the quantizer includes adaptive bit allocating means for allocating quantizing bits among the bands, the adaptive bit allocating means allocating no quantizing bits to bands at frequencies substantially higher than the cut-off frequency of the low-pass filter; and

the including means includes in the compressed digital signal no sub information for bands at frequencies substantially higher than the cut-off frequency of the low-pass filter.

10. The apparatus of claim 1, wherein the deriving means includes:

block dividing means for diving the digital input signal in time into blocks;

an orthogonal transform circuit, the orthogonal transform circuit receiving each block from the block dividing means, and providing a set of spectral coefficients in response to each block;

grouping means for grouping each set of spectral coefficients from the orthogonal transform circuit by frequency into the bands.

11. The apparatus of claim 10, wherein:

the apparatus additionally includes frequency range dividing means for dividing the digital input signal in frequency into a frequency range signal in each of plural frequency ranges;

the block dividing means is for dividing each frequency range signal in time into blocks; and

the orthogonal transform circuit includes means for orthogonally transforming each block of each frequency range signal to provide the spectral coefficients.

12. The apparatus of claim 11, wherein the frequency range dividing means divides the digital input signal in frequency into a frequency range signal in each of plural frequency ranges, each frequency range having a bandwidth, the frequency ranges including a lowest frequency range, and a next-lowest frequency range, the next-lowest frequency range having a bandwidth equal to the bandwidth of the lowest frequency range.

13. The apparatus of claim 11, wherein the frequency range dividing means divides the digital input signal in frequency into a frequency range signal in each of plural frequency ranges, the bandwidth of the frequency ranges increasing with increasing frequency.

14. The apparatus of claims 12 or 13, wherein the block dividing means divides the frequency range signal in each of the frequency ranges above the lowest frequency range into blocks, each block having a maximum block length, the maximum block lengths of the blocks being invariable between the compression modes.

15. The apparatus of claim 11, wherein the block dividing means divides each frequency range signal into blocks, each block having a block length, the block lengths of the blocks having a maximum block length depending on the selected one of the compression modes, the maximum block length being greater in the second compression mode than in the first compression mode.

16. The apparatus of claim 11 or 15, wherein the frequency range dividing means divides the digital input signal in frequency into a frequency range signal in each of plural frequency ranges, each frequency range having a bandwidth, the frequency ranges including a lowest frequency range, and a next-lowest frequency range, the next-lowest frequency range having a bandwidth equal to the bandwidth of the lowest frequency range.

17. The apparatus of claim 16, wherein the frequency range dividing means divides the digital input signal in frequency into a frequency range signal in each of plural frequency ranges, the bandwidth of the frequency ranges increasing with increasing frequency.

18. The apparatus of claim 16, wherein the block dividing means divides the frequency range signals in the frequency ranges above the lowest frequency range into blocks having a maximum block length, the maximum block length being invariable between the compression modes.

19. The apparatus of claim 11 or 15, wherein the apparatus is additionally for deriving an additional compressed digital signal, the additional compressed digital signal having a lower bit rate than the compressed digital signal, and wherein:

the apparatus additionally comprises inverse orthogonal transform means for inversely orthogonally transforming the spectral coefficients obtained by orthogonally transforming one of the frequency range signals to provide a restored frequency range signal;

additional block dividing means for dividing the restored frequency range signal in time into blocks, each block having a block length, the block lengths of the blocks having a maximum block length greater than the maximum block length whereinto the block dividing means divided the frequency range signal; and

additional orthogonal transform means for orthogonally transforming each block of the restored frequency range signal from the additional block dividing means.

20. The apparatus of claim 11, wherein the frequency range dividing means comprises:

a low-pass filter having a cut-off frequency set according to the selected one of the plural compression modes, the low-pass filter receiving the digital input signal, and providing a frequency-limited signal, the cut-off frequency of the low-pass filter being set to a higher frequency in the first compression mode than in the second compression mode; and

means for dividing the frequency-limited signal by frequency to provide the frequency range signal in each of the plural frequency ranges.

21. The apparatus of claim 20, wherein:

the quantizer includes adaptive bit allocating means for allocating quantizing bits among the bands, the adaptive bit allocating means allocating no quantizing bits to bands at frequencies substantially higher than the cut-off frequency of the low-pass filter; and

the including means includes in the compressed digital signal no sub information for bands at frequencies substantially higher than the cut-off frequency of the low-pass filter.

22. The apparatus of claim 10, wherein the orthogonal transform circuit includes means for performing a modified discrete cosine transform.

23. A method for compressing a digital input signal in a selected one of plural compression modes to provide a compressed digital signal, the plural compression modes including a first compression mode and a second compression mode, the compressed digital signal having a higher bit rate in the first compression mode than in the second compression mode, the method comprising the steps of:

deriving from the digital input signal spectral coefficients grouped by frequency and by time into bands, each band having a frequency width, the frequency width of at least one of the bands being set according to the selected one of the compression modes;

quantizing the spectral coefficients in each band to provide quantized spectral coefficients; and

including the quantized spectral coefficients in each band in the compressed digital signal, together with sub information for each band.

24. The method of claim 23, wherein, in the step of deriving spectral coefficients, the frequency width of the at least one band is set to a first frequency width in the first compression mode and is set to a second frequency width in the second compression mode, the second frequency width being greater than the first frequency width.

25. The method of claim 24, wherein the second frequency width is an integral number of times greater than the first frequency width.

26. The method of claims 24 or 25, wherein:

the digital input signal is a digital audio signal; and

in the step of deriving spectral coefficients, selected ones of the spectral coefficients are grouped by frequency into bands for determining quantizing noise and masking, and a majority of the bands have frequency widths set to be broader towards higher frequencies.

27. The method of claim 23, wherein:

the digital input signal is a digital audio signal; and

in the step of deriving spectral coefficients, selected ones of the spectral coefficients are grouped by frequency into bands for determining quantizing noise and masking, and a majority of the bands have frequency widths set to be broader towards higher frequencies.

28. The method of claim 27, wherein the step of deriving spectral coefficients includes the step of receiving the digital input signal at a fixed sampling frequency, the sampling frequency being invariable between the compression modes.

29. The method of claims 23 or 27, wherein the step of deriving spectral coefficients includes the steps of:

providing a low-pass filter having a cut-off frequency;

setting the cut-off frequency of the low-pass filter according to the selected one of the plural compression modes, the cut-off frequency being set to a higher frequency in the first compression mode than in the second compression mode;

filtering the digital input signal using the low-pass filter to provide a frequency-limited signal; and

deriving the spectral coefficients from the frequency-limited signal.

30. The method of claim 29, wherein the step of deriving spectral coefficients additionally includes the step of receiving the digital input signal a fixed sampling frequency, the sampling frequency being invariable between the compression modes.

31. The method of claim 29, wherein:

the step of quantizing the spectral coefficients includes the step of adaptively allocating quantizing bits among the bands, the quantizing bits being allocated such that no quantizing bits are allocated to bands at frequencies substantially higher than the cut-off frequency of the low-pass filter; and

in the step of including the quantized spectral coefficients and sub information in the compressed digital signal, no sub information is included in the compressed digital signal for bands at frequencies substantially higher than the cut-off frequency of the low-pass filter.

32. The method of claim 23, wherein the step of deriving spectral coefficients includes the steps of:

dividing the digital input signal in time into blocks;

orthogonally transforming each block of the digital input signal to provide a set of spectral coefficients; and

grouping the spectral coefficients in the set of spectral coefficients by frequency into the bands.

33. The method of claim 23, wherein the step of deriving spectral coefficients from the digital input signal includes the steps of:

dividing the digital input signal in frequency into a frequency range signal in each of plural frequency ranges;

dividing each frequency range signal in time into blocks;

orthogonally transforming blocks of the frequency range signals to provide the set of spectral coefficients; and

grouping the spectral coefficients in the set of spectral coefficients by frequency into the bands.

34. The method of claim 33, wherein, in the step of dividing the digital input signal in frequency, each of the frequency ranges has a bandwidth, and the digital input signal is divided into frequency range signals in frequency ranges including a lowest frequency range and a next-lowest frequency range, the next-lowest frequency range having a bandwidth equal to the bandwidth of the lowest frequency range.

35. The method of claim 33, wherein, in the step of dividing the digital input signal in frequency into frequency range signals in plural frequency ranges, the frequency ranges include a lowest frequency range, each of the frequency ranges has a bandwidth, and the bandwidth of the frequency ranges increase with increasing frequency.

36. The method of claims 33, 34, or 35, wherein, in the step of dividing each frequency range signal in time into blocks, the frequency range signal in each of the frequency ranges above the lowest frequency range is divided into blocks, each block having a maximum block length, the maximum block lengths of the blocks being invariable between the compression modes.

37. The method of claim 33, wherein, in the step of dividing each frequency range signal into blocks, each block has a block length, the block lengths of the blocks have a maximum block length, and each frequency range signal is divided into blocks having a maximum block length depending on the selected one of the compression modes, the maximum block length being greater in the second compression mode than in the first compression mode.

38. The method of claim 37, in the step of dividing the digital input signal in frequency, each of the frequency ranges has a bandwidth, and the digital input signal is divided into frequency range signals in frequency ranges including a lowest frequency range and a next-lowest frequency range, the next-lowest frequency range having a bandwidth equal to the bandwidth of the lowest frequency range.

39. The method of claim 37, wherein the frequency range dividing means divides the digital input signal in frequency into a frequency range signal in each of plural frequency ranges, the bandwidth of the frequency ranges increasing with increasing frequency.

40. The method of claims 37, 38, or 39, wherein, in the step of dividing each frequency range signal in time into blocks, the frequency range signals in the frequency ranges above the lowest frequency range are divided into blocks having a maximum block length, the maximum block length being invariable between the compression modes.

41. The method of claims 33 or 37 wherein the method is additionally for deriving an additional compressed digital signal, the additional compressed digital signal having a lower bit rate than the compressed digital signal, and wherein the method additionally comprises the steps of:

inversely orthogonally transforming the spectral coefficients provided orthogonally transforming one of the frequency range signals to provide a restored frequency range signal,

dividing the restored frequency range signal in time into blocks, each block having a block length, the block lengths of the blocks having a maximum block length greater than the maximum block length whereinto the step of dividing each frequency range signal into block divided the frequency range signal; and

orthogonally transforming each block of the restored frequency range signal from the additional block dividing means.

42. The method of claim 33, wherein the step of dividing the digital input signal in frequency includes the steps of:

providing a low-pass filter having a cut-off frequency;

setting the cut-off frequency of the low-pass filter according to the selected one of the plural compression modes, the cut-off frequency of the low-pass filter being set to a higher frequency in the first compression mode than in the second compression mode;

filtering the digital input signal using the low-pass filter to provide a frequency-limited signal; and

dividing the frequency-limited signal in frequency to provide the frequency range signal in each of the frequency ranges.

43. The method of claim 42, wherein:

the step of quantizing the spectral coefficients includes the step of adaptively allocating quantizing bits among the bands, the quantizing bits being allocated among the bands such that no quantizing bits are allocated to bands at frequencies substantially higher than the cut-off frequency of the low-pass filter; and

in the step of including the quantized spectral coefficient in the digital output signal, no sub information for bands at frequencies substantially higher than the cut-off frequency of the low-pass filter is included in the compressed digital signal.

44. The method of claims 23, 27, 33, or 42, additionally comprising the step of recording the compressed digital signal on a recording medium.

45. The method of claim 44, wherein, in the step of recording the compressed digital signal on a recording medium, compressed digital signal is recorded on a magneto-optical disc.

46. The method of claim 44, wherein, in the step of recording the compressed digital signal on a recording medium, compressed digital signal is recorded on a semiconductor recording medium.

47. The method of claim 44, wherein, in the step of recording the compressed digital signal on a recording medium, compressed digital signal is recorded on an IC memory card.

48. The method of claim 44, wherein, in the step of recording the compressed digital signal on a recording medium, compressed digital signal is recorded on an optical disc.

49. The method of claim 33, wherein, in the step of orthogonally transforming the digital input signal, the digital input signal is orthogonally transformed using a modified discrete cosine transform.

50. A digital signal processing system for compressing a digital input signal to provide a compressed digital signal and for expanding the compressed digital signal to provide a digital output signal, the system operating in a selected one of at least two different compression modes, including a first compression mode and a second compression mode, the compressed digital signal having a higher bit rate in the first compression mode than in the second compression mode, the system comprising:

a compressor, including;

compressed signal deriving means for analyzing the digital input signal into plural spectral coefficients grouped by frequency into plural bands, each band having a frequency width, for quantizing the spectral coefficients in each band to provide quantized spectral coefficients, and for providing the quantized spectral coefficients as the compressed digital signal, and

means for setting the frequency width of at least one of the bands whereinto the spectral coefficients are grouped by frequency, the frequency width being set according to the selected compression mode; and

an expander, including;

demultiplexing means for receiving the compressed digital signal and extracting the quantized spectral coefficients and the sub information therefrom,

dequantizing means for dequantizing the quantized spectral coefficients in each band using the sub information for the band to provide spectral coefficients, the frequency width of at least one of the bands being set according to the selected compression mode, and

output signal deriving means for deriving the digital output signal from the spectral coefficients.

51. The system of claim 50, wherein, in the compressor:

the digital input signal is a digital audio signal; and

the compressed signal deriving means groups selected ones of the spectral coefficients by frequency into bands for determining quantizing noise and masking, a majority of the bands having frequency widths set to be broader towards higher frequencies.

52. The system of claim 50, wherein:

in the compressor, the compressed signal deriving means includes;

block dividing means for diving the digital input signal in time into blocks,

an orthogonal transform circuit, the orthogonal transform circuit receiving each block of the digital input signal from the block dividing means, and providing a set of spectral coefficients in response to each block,

grouping means for grouping each set of spectral coefficients from the orthogonal transform circuit by frequency into the bands;

and

in the expander, the output signal deriving means includes an inverse orthogonal transform circuit, the inverse orthogonal transform circuit receiving the spectral coefficients from the dequantizing means and providing blocks of the digital output signal.

53. The apparatus of claim 52, wherein:

the compressor additionally includes frequency range dividing means for dividing the digital input signal in frequency into a frequency range signal in each of plural frequency ranges;

in the compressor;

the block dividing means is for dividing each frequency range signal in time into blocks, and

the orthogonal transform circuit includes means for orthogonally transforming each block of each frequency range signal to provide the spectral coefficients; and

in the expander;

the dequantizing means includes means for dividing the spectral coefficients by frequency into frequency ranges corresponding to the frequency ranges whereinto the frequency range dividing means divides the digital input signal,

the inverse orthogonal transform circuit includes means for inversely orthogonally transforming the spectral coefficients in each frequency range to provide blocks of a restored frequency range signal, and

the output signal deriving means additionally comprises means for synthesizing the blocks of the restored frequency range signals to provide the blocks of the digital output signal.

54. The apparatus of claim 53, wherein, in the compressor, the block dividing means divides each frequency range signal into blocks, each block having a block length, the block lengths of the blocks having a maximum block length depending on the selected one of the compression modes, the maximum block length being greater in the second compression mode than in the first compression mode.

55. The apparatus of claim 53, wherein, in the compressor the frequency range dividing means comprises:

a low-pass filter having a cut-off frequency set according to the selected one of the plural compression modes, the low-pass filter receiving the digital input signal, and providing a frequency-limited signal, the cut-off frequency of the low-pass filter being set to a higher frequency in the first compression mode than in the second compression mode; and

means for dividing the frequency-limited signal by frequency to provide the frequency range signal in each of the plural frequency ranges.

56. Method for deriving a second compressed digital signal from a first compressed digital signal obtained by compressing a digital input signal in a first compression mode, the second compressed digital signal being compressed in a second compression mode, the first compressed digital signal and the second compressed digital signal respectively having a first bit rate and a second bit rate, the second bit rate being less than the second bit rate, the first compressed digital signal including plural quantized spectral coefficients grouped by frequency into bands, the bands being grouped by frequency into frequency ranges, the frequency ranges including a lowest-frequency frequency range, the method comprising the steps of:

extracting the quantized spectral coefficients from the first compressed digital signal;

inversely quantizing the quantized spectral coefficients to provide recovered spectral coefficients;

inversely orthogonally transforming the recovered spectral coefficients in the lowest-frequency frequency range to provide a block of a lowest-frequency frequency range signal;

joining blocks of the lowest-frequency frequency range signal together to provide super blocks;

orthogonally transforming the super blocks of the lowest-frequency frequency range signal to provide new spectral coefficients;

re-quantizing the new spectral coefficients and the recovered spectral coefficients not inversely orthogonally transformed in the inverse orthogonal transform step to provide new quantized spectral coefficients; and

including the new quantized spectral coefficients in the second compressed digital signal.

57. The method of claim 56, wherein:

the frequency ranges additionally include a higher-frequency frequency range;

the method additionally comprises the steps of;

inversely orthogonally transforming the recovered spectral coefficients in a higher-frequency frequency range to provide a block of a higher-frequency frequency range signal,

downsampling the higher-frequency frequency range signal to provide a downsampled frequency range signal, and

orthogonally transforming the downsampled frequency range signal to provide additional new spectral coefficients; and

in the step of quantizing the new spectral coefficients and the recovered spectral coefficients not inversely orthogonally transformed in the inverse orthogonal transform step, the additional new spectral coefficients are also quantized to provide new quantized spectral coefficients.

58. The method of claim 57, wherein, in the step of downsampling the higher-frequency frequency range signal, the higher-frequency frequency range signal is downsampled by a factor of at least two.

59. The method of claim 56, additionally comprising the step of reproducing the first compressed signal from a recording medium.