System for subband coding of a digital audio signal and coder and decoder constituting the same

• US 5,105,463 A
• Filed: 11/27/1989
• Issued: 04/14/1992
• Est. Priority Date: 04/27/1987
• Status: Expired due to Fees

• Current Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Original Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Inventors: Breeuwer, Marcel, Veldhuis, Raymond N. J., van der Waal, Robbert G.
• Primary Examiner(s): Shaw, Dale M.
• Assistant Examiner(s): Knepper, David D.
• Application Number:
  US07/441,804
• Time in Patent Office: 869 Days
• Field of Search: 381/29-42, 364/513.5, 84/1.01, 375/25-27, 375/122
• US Class Current: 704/200.1
• CPC Class Codes: G10L 19/002 Dynamic bit allocation for perceptual audio coders G10L19/032
• CPC Class Codes: G10L 19/0204 using subband decomposition
• CPC Class Codes: G10L 25/18 the extracted parameters being spectral information of each sub-band
• CPC Class Codes: G11B 20/10527 Audio or video recording; Data buffering arrangements G11B20/12 - G11B20/18 take precedence
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect
• CPC Class Codes: H04B 1/667 using a division in frequency subbands for TV signals H04N19/63

Abstract

A system for subband coding of a digital audio signal x(k) includes in the coder (1) a filter bank (3) for splitting the audio signal band, with sampling rate reduction, into subbands (p=1, . . . P) of approximately critical bandwidth and in the decoder (2) a filter bank (5) for merging these subbands, with sampling rate increase. For each subband (p) the coder (1) comprises a detector (7(p)) for determining a parameter G(p;m) representative of the signal level in a block (p;m) of M samples of the subband signal x_p (k) as well as a quantizer (8(p)) for adaptively block quantizing this subband signal in response to parameter G(p;m), and the decoder (2) comprises a dequantizer (9(p)) for adaptively block dequantizing the quantized subband signal s_p (k) in response to parameter G(p;m). The quantizing characteristics are related to the noise-masking curve of the human auditory system, owing to which a high-quality of the replica x(k) of audio signal x(k) is attained with an average number of approximately 2.5 bits per sample for representing the output signals of the coder (1). The occasional audibility of quantizing noise in this replica x(k) is reduced effectively in that the coder (1) and decoder (2) contain identical bit allocation means (23, 24) responsive to a set of parameters G(p;m) for the higher group of subbands (p_im ≦p≦P) within an allocation window (FIG. 5) for allocating a number of B(p;m) bits per sample from a fixed predetermined number of B bits for this allocation window to the quantizer (8(p)) and the dequantizer (9(p)) for the block (p;m) of subband signal x_p (k) and s_p (k), respectively.

Claims

1. A coder for subband coding of a digitally sampled audio music signal, comprising:

(a) analysis filter means for dividing the frequency band of said signal into a number P of subbands, the successive subband numbers p(1<

p<

P) increasing with increasing frequency, the bandwidths of said subbands substantially corresponding to bandwidths which are critical to the human auditory system;

(b) means responsive to the digital signal samples in each subband for determining characteristic parameters G(p;

m) representative of the signal strength in respective blocks (p;

m) of a number M of such samples, m being the block number and M being the same for all blocks of all subbands; and

(c) means for adaptively quantizing each block of signal samples in each subband in accordance with the characteristic parameter G(p;

m) of such block, such that for subbands p≦

p_im, where p_im is a predetermined subband number greater than one, the signal samples in the blocks in such subbands are quantized with a fixed number of bits B(p), and for subbands p≧

p_im the signal samples in the blocks in such subbands are quantized with a variable number of bits B(p;

m); and

(d) bit allocation means for allocating said variable number of bits B(p;

m) from a fixed total number B of bits to said adaptive quantizing means for quantizing the blocks in the subbands p≧

p_im, such allocation comprising (i) allocation of no bits to blocks having characteristic parameters G(p;

m) which are below thresholds T(p) which correspond to thresholds of the human auditory system for perceiving just single music tones in such subbands, and (ii) allocation of bits to blocks having characteristic parameters G(p;

m) which exceed the thresholds T(p), the latter allocation being according to an algorithm which ensures that at least a minimum number of bits are allocated to each of such blocks and that at least one more than said minimum number of bits is allocated to the one of such blocks having the largest characteristic parameter G(p;

m).

2. A coder according to claim 1, wherein:

p_im is the number of a subband situated in the portion of the audio music frequency band having the lowest thresholds for masking noise in critical bands of the human auditory system by single music tones in the center of the respective critical bands.

3. A coder according to claim 2, wherein said bit allocation means comprises:

(e) comparator means for comparing said characteristic parameters G(p;

m) with said respective thresholds T(p) for subbands p≧

p_im, and for generating comparator signals C(p;

m) having a binary value of "1" when G(p;

m)≧

T(p) and having a binary value of "0" when G(p;

m)≦

T(p);

(f) means for storing a predetermined allocation pattern of said fixed number B(p) of quantizing bits for each of the subbands p≧

p_im, such allocation patterns corresponding to the threshold for masking noise in said critical bands of the human auditory system by single music tones in the center of such critical bands; and

(g) means for determining an allocation pattern of respective numbers B(p;

m) of quantizing bits for respective signal blocks in the subbands p≧

p_im, such allocation patterns being in accordance with said characteristic parameters G(p;

m) and said comparator signals C(p;

m).

4. A coder according to claim 3, wherein:

said digitally sampled audio music signal has a sampling rate 1/T;

said analysis filter means divides the frequency band of said signal into subbands by quadrature mirror filtering and sampling rate reduction; and

the bandwidth and signal sampling rate of each subband are respectively an integral submultiple of 1/2T and 1/T.

5. A coder according to claim 3, further comprising:

(h) storage means for storing the subband blocks and the characteristic parameters G(p;

m) applicable thereto;

and wherein said bit allocation means includes means for comparing the stored characteristic parameter G(p;

m) of a stored block (p;

m) of a subband p≧

p_im with the stored characteristic parameter G(p;

m+1) of the next stored block (p;

m+1) of the same subband, and based on such comparison allocates no quantizing bits to one of such blocks.

6. A coder according to claim 5, wherein:

said bit allocation means allocates no quantizing bits to the block (p;

m+1) when the ratio of G(p;

m)/G(p;

m+1) exceeds a predetermined ratio value R(p), and allocates no quantizing bits to the block (p;

m) when the ratio of G(p;

m)/G(p;

m+1) is smaller than the inverse 1/R(p) of said ratio value R(p).

7. A coder according to claim 6, wherein:

R(p) is of the order of one-hundred.

8. A coder according to claim 1, further comprising:

(e) storage means for storing said signal sample blocks (p;

m) and the characteristic parameters G(p;

m) applicable thereto;

and wherein said bit allocation means includes means for comparing the stored characteristic parameter G(p;

m) of a stored block (p;

m) of a subband p≧

p_im with the stored characteristic parameter G(p;

m+1) of the next stored block (p;

m+1) of the same subband, and based on such comparison allocates no quantizing bits to one of such blocks.

9. A coder according to claim 8, wherein:

said bit allocation means allocates no quantizing bits to said block (p;

m+1) when the ratio of G(p;

m)/G(p;

m+1) exceeds a predetermined ratio value R(p).

10. A coder according to claim 8, wherein:

said bit allocation means allocates no quantizing bits to said block (p;

m) when the ratio of G(p;

m)/G(p;

m+1) is smaller than a predetermined ratio value 1/R(p).

11. A coder according to claim 10, wherein:

said bit allocation means allocates no quantizing bits to said block (p;

m) when the ratio of G(p;

m)/G(p;

m+1) is smaller than the inverse 1/R(p) of said ratio value R(p).

12. A coder according to claim 11, wherein:

R(p) is of the order of one-hundred.

13. A decoder for decoding a digitally sampled audio music signal the frequency band of which has been divided into a number P of subbands, the successive subband number p(1≦

p≦

P) increasing with increasing frequency, each subband having successive blocks (p;

m) of the same number M of the signal samples in such subband, each such block having a characteristic parameter G(p;

m), the bandwidths of said subbands substantially corresponding to bandwidths which are critical to the human auditory system, the signal samples in each block having been quantized using numbers of bits determined in accordance with the characteristic parameter G(p;

m) of such block;

said decoder comprising;

(a) dequantizing means for receiving the quantized subband signals and the characteristic parameters G(p;

m) applicable thereto, said dequantizing means being adapted to dequantize each subband block in accordance with its characteristic parameter G(p;

m) and the number of bits used to quantize such block, the blocks in subbands p≦

p_im being dequantized by allocating a fixed number of bits B(p) thereto, and the blocks in subbands p≧

p_im being dequantized by allocating a variable number of bits B(p;

m) thereto, p_im being a predetermined subband number greater than one;

(b) bit allocation means for allocating said variable number of bits B(p;

m) to said dequantizing means for dequantizing the blocks in subbands p≧

p_im, said bit allocation means including

(i) means for receiving the respective characteristic parameters G(p;

m) of blocks (p;

m) of the subbands p≧

p_im ;

(ii) means for comparing the characteristic parameters G(p;

m) of said blocks with thresholds T(p) corresponding to thresholds of the human auditory system for perceiving just single music tones in the relevant subbands (p);

(iii) bit assignment means for assigning no bits to subband blocks having characteristic parameters G(p;

m) below the thresholds T(p) applicable thereto, and assigning from said fixed number of bits B(p) a number of bits B(p;

m) to subband blocks having characteristic parameters G(p;

m) which are equal to or greater than the applicable thresholds T(p), determination of said number of bits B(p;

m) being made according to an algorithm which assigns at least a minimum number of bits to each subband block having a characteristic parameter G(p;

m) which at least equals the applicable threshold T(p), the remainder of said fixed number of bits B(p) being assigned to the remaining subband blocks in accordance with their respective characteristic parameter values G(p;

m), the latter assignments being made in reverse order beginning with the subband block having the largest characteristic parameter value; and

(c) synthesis filter means for receiving the dequantized subband blocks produced by said dequantizing means and reconstructing a replica of the digitally sampled audio music signal by merging such subband blocks, such merger being effected by quadrature mirror filtering and increasing the sampling rate of the signal samples in such subblocks.

14. A decoder according to claim 13, further comprising:

(d) storage means for storing the signal sample blocks (p;

m) and the characteristic parameters G(p;

m) applicable thereto;

and wherein said bit assignment means comprises means for comparing the stored characteristic parameter G(p;

m) of a block (p;

n) of a subband p≧

p_im with the stored characteristic parameter G(p;

m+1) of the next block (p;

m+1) of the same subband, and based on such comparison assigns no dequantizing bits to one of the blocks (p;

m) and (p;

m+1).

15. A decoder according to claim 14, wherein:

said bit assignment means assigned no bits to the block (p;

m+1) when the ratio of G(p;

m)/G(p;

m+1) exceeds a predetermined ratio value R(p).

16. A decoder according to claim 14, wherein:

said means for assigning bits assigns no bits to the block (p;

m+1) when the ratio of G(p;

m)/G(p;

m+1) is smaller than a predetermined ratio value 1/R(p).

17. A decoder according to claim 15, wherein said bit assignment means assigns no bits to the block (p;

m) when the ratio of G(p;

m)/G(p;

m+1) is less than the inverse 1/R(p) of said ratio value R(p).

18. A decoder according to claim 17, wherein:

R(p) is of the order of one-hundred.