Optimization of MP3 audio encoding by scale factors and global quantization step size
View Patent ↗An iterative rate-distortion optimization algorithm for MPEG I/II Layer-3 (MP3) encoding based on the method of Lagrangian multipliers. Generally, an iterative method is performed such that a global quantization step size is determined while scale factors are fixed, and thereafter the scale factors are determined while the global quantization step size is fixed. This is repeated until a calculated rate-distortion cost is within a predetermined threshold. The methods are demonstrated to be computationally efficient and the resulting bit stream is fully standard compatible.
1. A method for encoding of an audio source sequence in an audio encoder, the encoding being dependent on quantization factors, the quantization factors including a global quantization step size and scale factors, the method comprising:
determining, using a processor, quantization factors that minimize a cost optimization function by iteratively;
selecting the global quantization step size to minimize the cost optimization function based on the scale factors, and
selecting the scale factors to minimize the cost optimization function based on the global quantization step size, wherein the scale factors are constrained within a bit length; and
encoding, using the processor, the audio source sequence based on the determined quantization factors.
2. The method claimed in claim 1 , wherein the scale factors include scale factor parameters scalefac, scalefac_compress, and scalefac_scale.
3. The method claimed in claim 2 , wherein the global quantization step size comprises a parameter global_gain.
4. The method claimed in claim 3 , further comprising calculating a value of scalefac which minimizes the cost function and constraining scalefac to within the bit length.
5. The method claimed in claim 4 , wherein calculating the value of scalefac includes differentially calculating the cost function with respect to scalefac to determine the value of scalefac which minimizes the cost function.
6. The method claimed in claim 1 , wherein the bit length is a first bit length for a first group of scale factor bands and the bit length is a second bit length for a second group of scale factor bands.
7. The method claimed in claim 1 , wherein the cost optimization function is a rate-distortion optimization function that includes a distortion term and a rate term.
8. The method claimed in claim 7 , wherein selecting the scale factors comprises fixing the global quantization step size and calculating the distortion term and the rate term for a plurality of combinations of the scale factors.
9. The method claimed in claim 7 , wherein the selecting the global quantization step size includes fixing the scale factors and calculating the distortion term and rate term for a plurality of global quantization step sizes.
10. The method claimed in claim 1 , wherein encoding comprises quantizing and entropy coding.
11. The method claimed in claim 1 , wherein the encoding is further dependent on quantized spectral coefficients, Huffman codebooks, and Huffman coding region partition, the method further comprising minimizing the cost function with respect to the quantized spectral coefficients, the Huffman codebooks, and the Huffman coding region partition.
12. An encoder device for encoding of an audio source sequence, the encoding being dependent on quantization factors, the quantization factors including a global quantization step size and scale factors, the encoder device comprising:
a processor;
a memory accessible by the processor; and
an encoding application stored in memory and executable by the processor, which, when executed, configures the processor to:
determine quantization factors that minimize a cost optimization function by iteratively
selecting the global quantization step size to minimize the cost optimization function based on the scale factors, and
selecting the scale factors to minimize the cost optimization function based on the global quantization step size, wherein the scale factors are constrained within a bit length; and
encode the audio source sequence based on the determined quantization factors.
13. The encoder device claimed in claim 12 , wherein the scale factors include scale factor parameters scalefac, scalefac_compress, and scalefac_scale.
14. The encoder device claimed in claim 13 , wherein the global quantization step size comprises a parameter global_gain.
15. The encoder device claimed in claim 14 , wherein the processor is further configured to calculate a value of scalefac which minimizes the cost function and to constrain scalefac to within the bit length.
16. The encoder device claimed in claim 15 , wherein the processor is configured to calculate the value of scalefac by differentially calculating the cost function with respect to scalefac to determine the value of scalefac which minimizes the cost function.
17. The encoder device claimed in claim 12 , wherein the bit length is a first bit length for a first group of scale factor bands and the bit length is a second bit length for a second group of scale factor bands.
18. The encoder device claimed in claim 12 , wherein the cost optimization function is a rate-distortion optimization function that includes a distortion term and a rate term.
19. The encoder device claimed in claim 18 , wherein the processor is configured to select the scale factors by fixing the global quantization step size and calculating the distortion term and the rate term for a plurality of combinations of the scale factors.
20. The encoder device claimed in claim 18 , wherein the processor is configured to select the global quantization step size by fixing the scale factors and calculating the distortion term and rate term for a plurality of global quantization step sizes.
21. The encoder device claimed in claim 12 , wherein the processor is configured to encode the audio source sequence by quantizing and entropy coding the sequence.
22. The encoder device claimed in claim 12 , wherein the processor is further configured to encode the source sequence based upon quantized spectral coefficients, Huffman codebooks, and a Huffman coding region partition, and wherein the processor is further configured to minimize the cost function with respect to the quantized spectral coefficients, the Huffman codebooks, and the Huffman coding region partition.