Scalable audio encoding / decoding method and apparatus
Patent Information
- Application Number
- MYPI1998002571
- Authority / Receiving Office
- MY · MY
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 1997-11-20
- Filing Date
- 1998-06-09
- Publication Date
- 1999-05-20
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Current audio encoding and decoding methods lack scalability and efficiency in processing audio signals, particularly in managing layered data streams effectively, which affects the quality and compression of audio data.
The method involves time/frequency mapping of audio signals, quantizing spectral data with a uniform scale factor across bands, packing data into bit streams, and using arithmetic coding for the base and enhancement layers, with side information and probability model details encoded sequentially from most significant to least significant bits.
This approach enables scalable and efficient audio encoding and decoding by ensuring consistent quality across layers, improving compression efficiency and maintaining audio fidelity.
Abstract
Description
A scalable audio encoding / decoding method and apparatus are provided. To code an audio signal into layered data streams having a base layer and at least one enhancement layer, the encoding method includes the steps of time / frequency mapping input audio signals and quantizing the spectral data with the same scale factor for each predetermined scalefactor band, and packing the quantized data into bit streams, wherein the bit stream generating step comprises the steps of coding the quantized data corresponding to the base layer, coding the quantized data corresponding to the next enhancement layer of the coded base layer and the remaining quantized data uncoded by a limit in a layer size and belonging to the coded layer, and performing the layer coding step for all enhancement layers to form bit streams. In the base layer coding step, the enhancement layer coding step and the sequential coding step, arithmetic coding is performed using a predetermined probability model in the order of bit sequences from the most significant bit sequence to the least significant bit sequence by representing the side information and quantized data corresponding to a layer to be coded in a predetermined number of bits. The side information contains scale factors and probability model information to be used in arithmetic coding.
Claims
CLAIMS 1. A scalable audio encoding method for coding audio signals into a layered data stream having a base layer and a plurality of enhancement layers, comprising the steps of quantizing input audio signals for each of a plurality of predetermined coding bands; and 5 generating bitstreams by packing the quantized data, wherein the bitstream generating step comprises: slicing said quantized data into sliced bits; coding sliced bits of the quantized data corresponding to the base layer; coding sliced bits of the quantized data corresponding to the first enhancement layer; and 10 sequentially coding sliced bits of the quantized data corresponding to each subsequent enhancement layer to form bitstreams, wherein each of the base layer coding step, the first enhancement layer coding step and the sequential enhancement coding steps include (i) coding both side information and quantized data corresponding to each respective layer to be represented by digits of a same 15 predetermined number; and (ii) arithmetic-coded using a predetermined probability model, and wherein the base layer coding step includes coding the most significant bits of the quantized data, the first enhancement layer coding step includes coding the next to most significant bits, and each subsequent enhancement coding steps includes coding progressively lower significant bits until the last enhancement layer includes coding the least 20 significant bits. -562. The scalable audio encoding method according to claim 1, wherein the step of coding side information includes coding scale factors and probability models to be used in the arithmeticcoding.
3. The scalable audio encoding method according to claim 2, wherein the step of coding the 5 scale factors comprises the steps of: obtaining the maximum scale factor; obtaining differences between the maximum scale factor and the respective scale factors; and arithmetic-coding the differences.
4. The scalable audio encoding method according to claim 3, wherein the step of arithmetic 10 coding the differences include using probability models listed in Tables 5.1 through 5.4.
5. The scalable audio encoding method according to claim 1, wherein the arithmetic coding step include using probability models listed in Tables 6.1 through 6.31.
6. The scalable audio encoding method according to claim 5, wherein the coding of the side information includes coding the probability models, and coding the probability models 15 includes the steps of: obtaining the minimum value of the probability model information values; obtaining differences between the minimum probability model information and the respective model information values; and arithmetic-coding the differences. -577. The scalable audio encoding method according to claim 6, wherein the arithmetic coding step includes using probability models listed in Tables 5.5 through 5.9.
8. A scalable audio encoding method according to claim 1, further comprising: receiving quantized data composed of sign data and magnitude data; 5 a. coding by a predetermined encoding method the most significant bit sequences composed of most significant bits of the magnitude data of the quantized data represented by the same number of bits; b. coding sign data corresponding to non-zero data among the coded most significant bit sequences; 10 c. coding the most significant bit sequences among uncoded magnitude data of the quantized data by said predetermined encoding method, d. coding uncoded sign data among the sign data corresponding to non-zero magnitude data among bit sequences; and e. performing the magnitude data coding step and the sign data coding step of steps c and d 15 on the respective bits of the quantized data until all of the quantized data is coded.
9. The scalable audio encoding method according to claim 8, wherein said predetermined encoding method includes arithmetic coding and said sign data coding steps include using a probability model having a size of 2 and cumulative frequency values of 8192 and 0.
10. The scalable audio encoding method according to claim 8, further comprising the step of 20 coupling respective bit sequences representing the coded magnitude data and sign data into units of bits of a predetermined number. -5811. The scalable audio encoding method according to claim 10, wherein said predetermined number of bits is 4.
12. The scalable audio encoding method according to claim 10, wherein a four-dimensional vector coupled in units of bits is divided into two subvectors according to the data pattern of 5 the two subvectors in coding the respective samples.
13. The scalable audio encoding method according to claim 1, wherein the bitrate of the base layer is 16 kbps and an interlayer bitrate is 1 kbps.
14. The scalable audio encoding method according to claim 1, further comprising coding a single set of header information commonly used for said base layer, first enhancement layer 10 and each subsequent enhancement layer.
15. The scalable audio encoding method according to claim 1, wherein the audio input signal quantization step includes the steps of: converting input audio signals of a temporal domain into signals of a frequency domain for each of a plurality of predetermined subbands; 15 coupling together the converted signals as signals of said predetermined subbands by time / frequency mapping; calculating a masking threshold at each predetermined subband; and quantizing the signals for each predetermined subband so that quantization noise of each band is smaller than the masking threshold. 20 16. The scalable audio encoding method according to claim 1, wherein the bitrate of the base layer is 16 kbps and an interlayer bitrate is 8 kbps. -5917. A scalable audio coding apparatus comprising: a quantizing portion (220) for (i) signal-processing input audio signals and (ii) quantizing the same for each coding band; and a bit packing portion (240) for (i) generating bitstreams by band-limiting a base layer into 5 scale factor bands so as to be scalable, (ii) coding side information corresponding to the base layer, (iii) coding the quantized information sequentially from the most significant bit sequence to the least significant bit sequence and from lower frequency components to higher frequency components corresponding to the base layer, and (iv) coding side information and the quantized data corresponding to the next enhancement layer, to perform 10 coding on all layers.
18. The scalable audio coding apparatus according to claim 17, wherein the quantizing portion (220) comprises: a time / frequency mapping portion (200) for converting the input audio signals of a temporal domain into signals of a frequency domain; 15 a psychoacoustic portion (210) for coupling the converted signals by signals of predetermined subbands by time / frequency mapping and calculating a masking threshold at each subband using a masking phenomenon generated by interaction of the respective signals; and a quantizing portion (220) for quantizing the signals for each predetermined coding band 20 while the quantization noise of each band is compared with the masking threshold.
19. A scalable audio decoding method for decoding audio data coded to have layered bitrates, comprising the steps of: -60decoding side information having at least scale factors and arithmetic-coding model information allotted to each band, said side information decoding step being repeated for each layer in the order of creation of the layers in datastreams having layered bitrates, by analyzing the significance of bits composing the datastreams, from upper significant bits to 5 lower significant bits, using the arithmetic coding models corresponding to the quantized data; restoring the decoded scale factors and quantized data into signals having the original magnitudes; and converting inversely quantized signals into signals of a temporal domain. 10 20. The scalable audio decoding method according to claim 19, wherein the decoding step decodes said side information in units of four-dimensional vectors.
21. The scalable audio decoding method according to claim 20, wherein the restoring step includes restoring four-dimensional vectors from two subvectors arithmetic-decoded according to the coding states of the respective samples. 15 22. The scalable audio decoding method according to claim 21, further comprising the step of arithmetic-decoding using the probability models listed in Tables 6.1 through 6.
31.
23. The scalable audio decoding method according to claim 19, wherein the decoding of the scale factors are performed by the steps of: decoding the maximum scale factor in the bitstream, arithmetic-decoding differences 20 between the maximum scale factor and the respective scale factors, and subtracting the differences from the maximum scale factor. -61 24. The scalable audio decoding method according to claim 23, further comprising the step of arithmetic-decoding the probability models listed in Tables 5.1 through 5.
4.
25. The scalable audio decoding method according to claim 23, further comprising arithmetic-decoding using the probability models listed in Tables 5.5 through 5.
8. 5 26. The scalable audio decoding method according to claim 19, wherein the decoding of arithmetic model indices is performed by decoding the minimum arithmetic model index in the bitstream, decoding differences between the minimum index and the respective indices in the side information of the respective layers, and adding the minimum index and the differences. 10 27. A computer usable medium, tangibly embodying a program of instructions executable by the machine to perform a scalable audio coding method for coding audio signals into a layered data stream having a base layer and enhancement layers of a predetermined number, the program comprising the steps of: quantizing input audio signals for each of a plurality of predetermined coding bands; and 15 generating bitstreams by packing the quantized data, wherein the bitstream generating step comprises: coding the quantized data corresponding to the base layer; coding the quantized data corresponding to the first enhancement layer of the coded base layer; and20 sequentially coding sliced bits of the quantized data corresponding to each subsequent enhancement layer to form bitstreams, -62wherein each of the base layer coding step, the first enhancement layer coding step and the sequential enhancement coding steps include (i) coding both side information and quantized data corresponding to each respective layer to be represented by digits of a same predetermined number; and (ii) arithmetic-coded using a predetermined probability model, 5 and wherein the base layer coding step includes coding the most significant bits of the quantized data, the first enhancement layer coding step includes coding the next to most significant bits, and each subsequent enhancement coding step includes coding progressively lower significant bits until the last enhancement layer includes coding the least significant bits. 10 28. The program storage device according to claim 27, wherein the scale factor coding step comprises the steps of obtaining the maximum scale factor; obtaining differences between the maximum scale factor and the respective scale factors; and arithmetic-coding the same. 15 29. The program storage device according to claim 28, wherein the coding of the side information includes coding the probability models and coding the probability models includes the steps of: obtaining the minimum value of the probability model information values; obtaining differences between the minimum probability model information and the respective 20 model information values; and arithmetic-coding the differences using the probability models listed in Tables 5.5 through 5.
9. -63 30. A computer usable medium, tangibly embodying a program of instructions executable by the machine to perform a scalable audio decoding method for decoding audio data coded to have layered bitrates, comprising the steps of: receiving and decoding side information having at least scale factors and arithmetic-coding 5 model information allotted to each band, said side information decoding step being repeated for each layer in the order of creation of the layers in datastreams having layered bitrates, by analyzing the significance of bits composing the datastreams, from upper significant bits to lower significant bits, using the arithmetic coding models corresponding to the quantized data; 10 restoring the decoded scale factors and quantized data into signals having the original magnitudes; and converting inversely quantized signals into signals of a temporal domain, a recording medium capable of reading a program for executing the scalable audio encoding method using a computer. 15 31. The program storage device according to claim 30, wherein said decoding step includes decoding side information in units of four-dimensional vectors, and decoding bit-sliced information of four samples in the four-dimensional vectors.
32. The program storage device according to claim 30, wherein the decoding of the scale factors is performed by decoding the maximum scale factor, arithmetic-coding the 20 differences between the maximum scale factor and the respective scale factors and subtracting the differences from the maximum scale factor.
33. The program storage device according to claim 30, wherein the decoding of the arithmetic model indices is performed by decoding the minimum arithmetic model index in -64the bitstream, decoding differences between the minimum index and the respective indices in the side information of the respective layers, and adding the minimum index and the differences.
34. The scalable audio encoding method according to claim 11, wherein a four-dimensional 5 vector coupled in units of bits is divided into two subvectors according to its data patterns in coding the respective samples.
35. A scalable audio encoding method for coding audio signals into a layered data stream having a base layer and enhancement layers of a predetermined number, comprising the steps of: 10 signal-processing input audio signals and quantizing the same for each predetermined coding band; and packing the quantized data to generate bitstreams, wherein the bitstream generating step comprises: coding the quantized data of a first range of frequencies corresponding to the base layer; 15 coding the quantized data of a second range of frequencies corresponding to a first enhancement layer; and sequentially coding sliced bits of the quantized data corresponding to each subsequent enhancement layer to form bitstreams, wherein each of the base layer coding step, the first enhancement layer coding step and the 20 sequential enhancement coding steps include (i) coding both side information and quantized data corresponding to each respective layer to be represented by digits of a same predetermined number; and (ii) arithmetic-coded using a predetermined probability model, -65 and wherein the base layer coding step includes coding the most significant bits of the quantized data, the first enhancement layer coding step includes coding the next to most significant bits, and each subsequent enhancement coding step includes coding progressively lower significant bits until the last enhancement layer includes coding the least significant 5 bits.
36. The scalable audio coding method according to claim 35, said second range of frequencies includes frequencies different from said first range of frequencies wherein the most significant digits of the quantized data of said frequencies different from said first range of frequencies are coded before coding next significant digits of the quantized data belonging 10 to the corresponding enhancement layer.
37. The scalable audio encoding method according to claim 36, wherein the step of coding the scale factors comprises the steps of: obtaining the maximum scale factor; and obtaining differences between the maximum scale factor and the respective scale factors and 15 arithmetic-coding the differences.
38. The scalable audio encoding method according to claim 37, wherein the probability models listed in Tables 5.1 through 5.4 are used in the step of arithmetic-coding the differences.
39. The scalable audio encoding method according to claim 36, wherein the probability 20 models listed in Tables 6.1 through 6.31 are used in the arithmetic-coding step.
40. The scalable audio encoding method according to claim 39, wherein the coding of the information for the probability models is performed by the steps of: -66obtaining the minimum value of the probability model information values; obtaining differences between the minimum probability model information and the respective model information values and arithmetic-coding the differences.
41. The scalable audio encoding method according to claim 40, wherein the probability 5 models listed in Tables 5.5 through 5.9 are used in the arithmetic-coding step.
42. The scalable audio encoding method according to claim 36, wherein, when the received quantized data is composed of sign data and magnitude data, the coding step comprises the steps of: a. coding by a predetermined encoding method the most significant bit sequences composed 10 of most significant bits of the magnitude data of the quantized data represented by the same number of bits; b. coding sign data corresponding to non-zero data among the coded most significant bit sequences; c. coding the most significant bit sequences among uncoded magnitude data of the quantized 15 data by said predetermined encoding method; d. coding uncoded sign data among the sign data corresponding to non-zero magnitude data among bit sequences; and e. performing the magnitude data coding step and the sign data coding step on the respective bits of the quantized data until all of the quantized data is coded. -67 43. The scalable audio encoding method according to claim 42, wherein said predetermined encoding method includes arithmetic coding and sign data coding steps include using a probability model having a size of 2 and cumulative frequency values of 8192 and 0.
44. The scalable audio encoding method according to claim 42, further comprising the step of 5 coupling respective bit sequences representing the coded magnitude data and sign data, into units of bits of a predetermined number.
45. The scalable audio encoding method according to claim 44, wherein said predetermined number of bits is 4.
46. The scalable audio encoding method according to claim 44, wherein a four-dimensional 10 vector coupled in units of bits is divided into two subvectors according to its data patterns in coding the respective samples.
47. The scalable audio encoding method according to claim 45, wherein a four-dimensional vector coupled in units of bits is divided into two subvectors according to its data patterns in coding the respective samples. 15 48. The scalable audio encoding method according to claim 36, wherein the bitrate of the base layer is 16 kbps and an interlayer bitrate is 8 kbps. V 49. The scalable audio encoding method according to claim 36, wherein the bitrate of the base layer is 16 kbps and an interlayer bitrate is 1 kbps.
50. The scalable audio encoding method according to claim 36, further comprising coding a 20 single set of header information commonly used for said base layer, first enhancement layer and each subsequent enhancement layer. -68 51. The scalable audio encoding method according to claim 36, wherein the audio input signal quantization includes the steps of: coverting input audio signals of a temporal domain into signals of a frequency domain for each of a plurality of predetermined subbands; 5 coupling together the converted signals as signals of said predetermined subbands by time / frequency mapping; calculating a masking threshold at each subband; and quantizing the signals for each predetermined subband so that quantization noise of each band is smaller than the masking threshold. 10 52. A scalable audio decoding apparatus comprising: a bitstream analyzing portion (400) for receiving and decoding side information having at least scale factors and arithmetic model information and quantized data, in the order of creation of the layers in layered bitstreams; an inverse quantizing portion (410) for restoring the decoded scale factors and quantized data 15 into signals having the original magnitudes; and a frequency / time mapping portion (420) for converting inversely quantized signals of a frequency domain into signals of a temporal domain to be output. -69ό I-1 fj-l -70FIG. 2 I_____________________I FIG. 3
