AUDIO PROCESSING

• US 20110096933A1
• Filed: 03/10/2009
• Published: 04/28/2011
• Est. Priority Date: 03/11/2008
• Status: Active Grant

• Current Assignee: Oxford Digital Limited
• Original Assignee: Oxford Digital Limited
• Inventors: Eastty, Peter Charles
• Granted Patent: US 9,203,366 B2
• Time in Patent Office: Days
• US Class Current: 381/56
• CPC Class Codes: H03G 3/00 Gain control in amplifiers or frequency changers
• CPC Class Codes: H03G 5/005 of digital signals
• CPC Class Codes: H03H 17/0294 Variable filters; Programmable filters

Abstract

A method of generating a desired shelving filter for audio data, the desired shelving filter having a low-corner frequency U and a high-corner frequency f₂, wherein the difference between the gain of the frequency response of the desired shelving filter at f₂ and the gain of the frequency response of the desired shelving filter at U is substantially equal to a non-zero desired amount, the method comprising: determining an order for a first shelving filter of a first predetermined type, wherein the low-corner frequency of the first shelving filter is U and the high-corner frequency of the first shelving filter is fΣ, determining, for a second shelving filter of a second predetermined type and of a predetermined order m, a frequency fz for the low-corner frequency of the second shelving filter and a frequency Z₄ for the high-corner frequency of the second shelving filter, where h is at least fi, f₄ is greater than f₃, and f₄ is at most fi, and forming the desired shelving filter as a filter combination of the first shelving filter and the second shelving filter; wherein fz, f₄ and the order of the first shelving filter are determined such that difference between the gain of the frequency response of the filter combination at fe and the gain of the frequency response of the filter combination at fi is substantially equal to the desired amount.

Claims

1. A method of generating a desired shelving filter for audio data, the desired shelving filter having a low-corner frequency f₁ and a high-corner frequency f₂, wherein the difference between the gain of the frequency response of the desired shelving filter at f₂ and the gain of the frequency response of the desired shelving filter at f₁ is substantially equal to a non-zero desired amount, the method comprising:

determining an order for a first shelving filter of a first predetermined type, wherein the low-corner frequency of the first shelving filter is f₁ and the high-corner frequency of the first shelving filter is f₂;

determining, for a second shelving filter of a second predetermined type and of a predetermined order m, a frequency f₃ for the low-corner frequency of the second shelving filter and a frequency f₄ for the high-corner frequency of the second shelving filter, where f₃ is at least f₄ is greater than f₃, and f₄ is at most f₂; and

forming the desired shelving filter as a filter combination of the first shelving filter and the second shelving filter;

wherein f₃, f₄ and the order of the first shelving filter are determined such that difference between the gain of the frequency response of the filter combination at f₂ and the gain of the frequency response of the filter combination at f₁ is substantially equal to the desired amount.

3. A method according to any one of the preceding claims, in which determining the order of the first shelving filter comprises determining the greatest order for the first shelving filter for which the magnitude of the desired amount is greater than the magnitude of the difference between the gain of the frequency response of the first shelving filter at f₂ and the gain of the frequency response of the first shelving filter at f₁.

4. A method according to claim 1 or 2, in which determining the order of the first shelving filter comprises determining the least order for the first shelving filter for which the magnitude of the desired amount is less than the magnitude of the difference between the gain of the frequency response of the first shelving filter at f₂ and the gain of the frequency response of the first shelving filter at f₁.

5. A method according to any one of the preceding claims, in which:

if the desired amount is positive, then the first shelving filter is arranged such that the gain of the frequency response of the first shelving filter at f₂ is greater than the gain of the frequency response of the first shelving filter at f₁; and

if the desired amount is negative, then the first shelving filter is arranged such that the gain of the frequency response of the first shelving filter at f₂ is less than the gain of the frequency response of the first shelving filter at f₁.

6. A method according to any one of the preceding claims, in which determining f₃ and f₄ comprises:

setting f₃ and f₄ relative to each other such that the difference between the gain of the frequency response of the second shelving filter at f₄ and the gain of the frequency response of the second shelving filter at f₃ is substantially equal to the difference between the desired amount and the difference between the gain of the frequency response of the first shelving filter at f₂ and the gain of the frequency response of the first shelving filter at f₁.

7. A method according to claim 6, in which f₃ and f₄ are set so that the range of frequencies from f₃ to f₄ is substantially centred in the range of frequencies from f₁ to f₂.

8. A method according to claim 7, in which f₃ and f₄ are set so that f₄+f₃=f₂+f₁ or f₄ f₃=f₂f₁.

11. A method according to any one of the preceding claims, comprising the step of a user specifying f₁, f₂ and the desired amount.

12. A method according to any one of the preceding claims, in which the first predetermined type and the second predetermined type are each filter types based on properties of a respective one of:

Butterworth low-pass filters;

Chebychev low-pass filters; and

elliptical low-pass filters.

13. A method according to any one of the preceding claims, in which the first predetermined type is the same as the second predetermined type.

18. A method according to claim 17, comprising:

for each frequency response indicator, maintaining a corresponding target gain;

in which, for a pair of adjacent frequency response indicators comprising a first frequency response indicator specifying a frequency f_A and an adjacent second frequency response indicator specifying a frequency f_B greater than f_A, the step of generating a shelving filter comprises carrying out a method according to any one of claims 1 to 13 in which;

f₁ is based on f_A;

f₂ is based on f_B; and

the desired amount is set to be the difference between the target gain corresponding to the second frequency response indicator and the target gain corresponding to the first frequency response indicator.

19. A method according to claim 18, comprising initialising each target gain to the gain specified by the corresponding frequency response indicator.

20. A method according to claim 18 or 19, comprising, for a pair of adjacent frequency response indicators comprising a first frequency response indicator and an adjacent second frequency response indicator:

before generating the shelving filter corresponding to that pair of adjacent frequency response indicators, determining whether the order of that shelving filter will exceed a predetermined maximum threshold and, if so, adjusting one or both of the target gain corresponding to the first frequency response indicator and the target gain corresponding to the second frequency response indicator to reduce the order of that shelving filter.

21. A method according to any one of claims 18 to 20, in which, for a pair of adjacent frequency response indicators comprising a first frequency response indicator specifying a frequency f_A and an adjacent second frequency response indicator specifying a frequency f_B greater than f_A, the step of carrying out a method according to any one of claims 1 to 13 to generate the shelving filter corresponding to that pair of adjacent frequency response indicators comprises setting f₁ equal to f_A and setting f₂ equal to f_B.

22. A method according to any one of claims 18 to 20, comprising:

for each pair of adjacent frequency response indicators, maintaining a corresponding frequency-proportion value;

in which, for a pair of adjacent frequency response indicators comprising a first frequency response indicator specifying a frequency f_A and an adjacent second frequency response indicator specifying a frequency f_B greater than f_A, the step of carrying out a method according to any one of claims 1 to 13 to generate the shelving filter corresponding to that pair of adjacent frequency response indicators comprises determining f₁ and f₂ based on f_A, f_B and the corresponding frequency-proportion value.

23. A method according to claim 22, in which f₂−

f₁=h(f_B−

f_A), where h is the frequency-proportion value corresponding to that pair of adjacent frequency response indicators.

24. A method according to claim 22, in which f₂/f₁=f_B/f_A)^h, where h is the frequency-proportion value corresponding to that pair of adjacent frequency response indicators.

25. A method according to any one of claims 22 to 24, in which the range from f₁ to f₂ is substantially centred relative to the range from f_A to f_B.

26. A method according to claim 25, when dependent on claim 23, in which f₂=f₁=f_B+f_A.

27. A method according to claim 25, when dependent on claim 24, in which f₂f₁=f_Af_B.

28. A method according to any one of claims 22 to 27, comprising, for a pair of adjacent frequency response indicators comprising a first frequency response indicator and an adjacent second frequency response indicator:

before generating the shelving filter corresponding to that pair of adjacent frequency response indicators, determining whether the order of that shelving filter will exceed a predetermined maximum threshold and, if so, increasing the frequency-proportion value corresponding to that pair of adjacent frequency response indicators to reduce the order of that shelving filter.

29. A method according to claim 28, in which the frequency-proportion value may not be increased above a predetermined maximum frequency-proportion value.

30. A method according to any one of claims 22 to 29, comprising initialising each frequency proportion value to a corresponding initialisation value.

31. A method according to claim 30, in which the initialisation value is in the range from about ⅓

to about ½

.

32. A method according to any one of claims 18 to 31, comprising;

after generating the set of one or more filters, performing one or more iterations, wherein an iteration comprises;

adjusting one or more of the target gains; and

repeating the step of generating using the adjusted target gains to update the set of one or more filters;

wherein the step of adjusting comprises adjusting one or more of the target gains so that an error between the desired frequency response and the frequency response of a combination of the one or more filters in the updated set of filters is less than an error between the desired frequency response and the frequency response of a combination of the one or more filters in the set of filters prior being updated.

33. A method according to claim 32, in which performing one or more iterations comprises performing at most a predetermined maximum number of iterations.

34. A method according to claim 32 or 33, in which an iteration comprises determining an error between the desired frequency response and the frequency response of a combination of the one or more filters in the set of filters, wherein the step of performing one or more iterations is terminated if the determined error is less than a predetermined error threshold.

35. A method according to claim 34, in which determining an error comprises:

for each of the frequency response indicators, determining the difference between the gain of the desired frequency response and the gain of the frequency response of the combination of the one or more filters in the set of filters at the frequency specified by that frequency response indicator; and

determining the error based on the determined differences.

36. A method according to any one of claims 32 to 35, in which adjusting a target gain corresponding to a frequency response indicator comprises subtracting, from that target gain, the difference between the gain of the frequency response of the combination of the filters in the set of filters at the frequency specified by that frequency response indicator and the gain specified by that frequency response indicator.

41. A method of processing audio data, the method comprising:

filtering the audio data by applying a first desired filter to the audio data, the first desired filter being a filter generated by a method according to any one of claims 1 to 38.

42. A method according to claim 41, comprising:

changing the filtering of the audio data from applying the first desired filter to the audio data to applying a second desired filter to the audio data, the second desired filter being a filter generated by a method according to any one of claims 1 to 38.

45. A method of configuring a processor for processing audio data, the method comprising:

generating a desired filter for the audio data by carrying out a method according to any one of claims 1 to 40; and

configuring the processor to process the audio data using the generated filter.

46. An apparatus for generating a desired audio filter, the apparatus comprising a processor arranged to generate a desired filter for audio data by carrying out a method according to any one of claims 1 to 40.

47. An apparatus for processing audio data, the apparatus comprising:

a processor arranged to process audio data by carrying out a method according to any one of claims 41 to 45;

means for supplying audio data to the processor for processing.

48. An apparatus according to claim 47, in which the processor is arranged to generate a desired filter for processing the audio data by carrying out a method according to any one of claims 1 to 40.

49. An audio filter generated by a method according to any one of claims 1 to 40.

50. A computer program which, when executed by a computer, carries out a method according to any one of claims 1 to 45.

51. A data carrying medium carrying a computer program according to claim 50.

52. A medium according to claim 51, in which the medium is a storage medium or a transmission medium.

2. A method according to claim 2, in which m=1.

9. A method according to claim 2, in which:

the desired amount is positive;

each zero of the transfer function of the first shelving filter is located at a respective location determined by interpolating between (a) the location of a corresponding pole of the transfer function of a first low-pass filter of a type corresponding to the first predetermined type and having a cut-off frequency of f₁ and an order of n−

1 and (b) the location of a corresponding pole of the transfer function of a second low-pass filter of that corresponding type and having a cut-off frequency of f₁ and an order of n; and

each pole of the transfer function of the first shelving filter is located at a respective location determined by interpolating between (c) the location of a corresponding pole of the transfer function of a third low-pass filter of that corresponding type and having a cut-off frequency of f₂ and an order of n−

1 and (d) the location of a corresponding pole of the transfer function for a fourth low-pass filter of that corresponding type and having a cut-off frequency of f₂ and an order of n.

10. A method according to claim 2, in which:

the desired amount is negative;

each pole of the transfer function of the first shelving filter is located at a respective location determined by interpolating between (a) the location of a corresponding pole of the transfer function of a first low-pass filter of a type corresponding to the first predetermined type and having a cut-off frequency of f₁ and an order of n−

1 and (b) the location of a corresponding pole of the transfer function of a second low-pass filter of that corresponding type and having a cut-off frequency of f₁ and an order of n; and

each zero of the transfer function of the first shelving filter is located at a respective location determined by interpolating between (c) the location of a corresponding pole of the transfer function of a third low-pass filter of that corresponding type and having a cut-off frequency of f₂ and an order of n−

1 and (d) the location of a corresponding pole of the transfer function for a fourth low-pass filter of that corresponding type and having a cut-off frequency of f₂ and an order of n.

14. A method of generating a desired filter for audio data, the method comprising:

specifying a plurality of frequency response indicators to define a desired frequency response for the desired filter, each frequency response indicator specifying a user-defined frequency and a corresponding user-defined gain for the desired frequency response at that frequency; and

determining a set of one or more filters such that, for each of the frequency response indicators, the gain of the frequency response of a combination of the one or more filters in the set of filters at the frequency specified by that frequency response indicator is substantially equal to the gain specified by that frequency response indicator.

15. A method according to claim 14, in which the number of frequency response indicators is user-defined.

16. A method according to claim 14 or 15, comprising:

the user defining one or more of the plurality of frequency response indicators.

17. A method according to any one of claims 14 to 16, in which determining the set of one or more filters comprises:

for each pair of adjacent frequency response indicators, generating a corresponding shelving filter, wherein a frequency response indicator specifying a frequency f_A and another frequency response indicator specifying a frequency f_B greater than f_A are adjacent if none of the plurality of frequency response indicators specifies a frequency between f_A and f_B.

37. A method according to any one of claims 14 to 36, comprising:

a user specifying a gain scale factor; and

for each of the frequency response indicators, scaling the gain specified by that frequency response indicator by the specified gain scale factor.

43. A method according to claim 42, in which the first desired filter and the second desired filter are generated by a method according to claim 14, the method comprising:

specifying a plurality of frequency response indicators to generate the first desired filter;

updating the plurality of frequency response indicators; and

generating the second desired filter based on the updated plurality of frequency response indicators.

38. A method of implementing an audio filter, in which the transfer function of the audio filter comprises 2s complex-valued zeros and 2s complex-valued poles, the method comprising:

forming s second-order-filter-sections, such that, for each integer i in the range the 1≦

i≦

s, the i-th second-order-filter-section is based on;

(a) the complex-valued zero of the transfer function of the audio filter that, together with the associated complex conjugate zero, is the i-th closest pair of complex-valued zero and associated complex conjugate zero of the transfer function of the audio filter to the unit circle; and

(b) the complex-valued pole of the transfer function of the audio filter that, together with the associated complex conjugate pole, is the i-th closest pair of complex-valued pole and associated complex conjugate pole of the transfer function of the audio filter to the unit circle.

39. A method of changing from filtering audio data with a first audio filter to filtering the audio data with a second audio filter, in which the transfer function of the first audio filter comprises 2s complex-valued zeros and 2s complex-valued poles, and in which the transfer function of the second audio filter comprises 2t complex-valued zeros and 2t complex-valued poles, the method comprising:

forming the first filter using s second-order-filter-sections, in which, for each integer i in the range 1≦

i≦

s, the i-th second-order-filter-section is based on (a) the complex-valued zero of the transfer function of the first audio filter that, together with the associated complex conjugate zero, is the i-th closest pair of complex-valued zero and associated complex conjugate zero of the transfer function of the first audio filter to the unit circle; and

(b) the complex-valued pole of the transfer function of the first audio filter that, together with the associated complex conjugate pole, is the i-th closest pair of complex-valued pole and associated complex conjugate pole of the transfer function of the first audio filter to the unit circle;

forming the second filter using t second-order-filter-sections, in which, for each integer j in the range 1≦

j≦

t, the j-th second-order-filter-section is based on (c) the complex-valued zero of the transfer function of the second audio filter that, together with the associated complex conjugate zero, is the j-th closest pair of complex-valued zero and associated complex conjugate zero of the transfer function of the second audio filter to the unit circle; and

(d) the complex-valued pole of the transfer function of the second audio filter that, together with the associated complex conjugate pole, is the j-th closest pair of complex-valued pole and associated complex conjugate pole of the transfer function of the second audio filter to the unit circle; and

for each integer k in the range 1≦

k≦

minimum(s,t), setting the value of a state-variable for the k-th second-order-filter-section of the second audio filter to be the value of a corresponding state-variable for the k-th second-order-filter-section of the first audio filter.

40. A method according to claim 39, comprising:

if t is greater than s, then, for each integer r in the range s+1≦

r≦

t;

initialising the state variables for the r-th second-order-filter section of the second audio filter to the value 0;

providing audio input to the r-th second-order-filter section of the second audio filter for a predetermined period of time after beginning to change from filtering the audio data with the first audio filter to filtering the audio data with the second audio filter; and

after the predetermined period of time, starting to use the r-th second-order-filter section of the second audio filter to filter the audio data.

44. A method according to claim 42 or 43, in which the step of changing comprises carrying out a method according to claim 39 or 40.