Room acoustic response modeling and equalization with linear predictive coding and parametric filters

• US 8,363,853 B2
• Filed: 02/23/2007
• Issued: 01/29/2013
• Est. Priority Date: 02/23/2007
• Status: Active Grant

• Current Assignee: Audyssey Laboratories, Inc.
• Original Assignee: Audyssey Laboratories, Inc.
• Inventors: Bharitkar, Sunil, Zhang, Yun, Kyriakakis, Chris
• Primary Examiner(s): Chin, Vivian
• Assistant Examiner(s): Zhang, Leshui
• Application Number:
  US11/710,019
• Publication Number:
  US 20080205667A1
• Time in Patent Office: 2,167 Days
• Field of Search: 381/66, 381 711- 7112, 381/92, 381/93, 381/94.1, 381/94.2, 381/94.3, 381/94.4, 381/94.7, 381/95, 381/98, 381/97, 381/100, 381/103, 381/5, 381/9, 381/61, 381/63, 381/56, 381/58, 381/71, 381/73.1, 381/30, 381/94, 381/106, 37940601-40616, 375/344, 375/319, 375/331, 375/349, 375/350, 375/229, 375/232, 375/234
• US Class Current: 381/103
• CPC Class Codes: H04R 3/00 Circuits for transducers , loudspeakers or microphones

Abstract

A method for determining coefficients of a family of cascaded second order Infinite Impulse Response (IIR) parametric filters used for equalizing a room response. The method includes determining parameters of each IIR parametric filter from poles or roots of a reasonably high-order Linear Predictive Coding (LPC) model. The LPC model is able to accurately model the low-frequency room response modes providing better equalization of loudspeaker and room acoustics, particularly at the low frequencies. Advantages of the method include fast and efficient computation of the LPC model using a Levinson-Durbin recursion to solve the normal equations that arise from the least squares formulation. Due to possible band interactions between the cascaded IIR parametric filters, the method further includes optimizing the Q value of each filter to better equalize the room response.

Claims

1. A method for equalizing audio signals provided to speakers, the method comprising:

measuring loudspeaker-room acoustics to obtain time domain room response data;

processing the time domain room response data with a Linear Predictive Coding (LPC) model to obtain smoothed time domain room response data;

performing an FFT on the smoothed time domain room response data to obtain smoothed frequency domain room response data;

computing center frequency F_c, a gain G, a bandwidth BW, and a term Q factor for a plurality of parametric Infinite-duration Impulse Response (IIR) filters from the smoothed frequency domain room response data includes a method for determining a number of peaks num_peaks comprising;

computing a frequency response HH2 from a linear predication coefficient q wherein elements in the frequency response HH2 correspond to an array of frequencies FF;

given an interested frequency range between LO_FREQ and HI_FREQ in Hz, and a low bin bin_lo and a high bin, bin_hi, updating the array of frequencies FF such that the elements of the updated the array of frequencies FF are the frequencies in the interested frequency range;

computing HH2_abs, the magnitude of HH2, based on the bin_lo and bin_hi in the interested frequency range;

determining peak locations peak_loc and valley_locations valley_loc while ensuring that a first peak occurs before a first valley;

saving a number of peak locations as the num_peaks; and

determining the center frequency Fc and gain G of each peak based on the peak location peak_loc and the magnitude HH2_abs at the peak location;

cascading the plurality of parametric IIR filters to form an equalizing filter;

equalizing a signal with the equalization filter to obtain an improved loudspeaker-room response; and

providing the equalized signal to a speaker to provide improved sound production.

2. The method of claim 1, wherein processing the time domain room response data with an LPC model includes processing the time domain room response data using a Levinson-Durbin recursion.

3. The method of claim 2, wherein processing the time domain room response data with an LPC model comprises processing the time domain room response data with a high-order LPC model.

4. The method of claim 1 further including optimizing the Q term and gain term of each of the plurality of parametric IIR filters.

5. The method of claim 4, wherein optimizing the Q term of each of the plurality of parametric IIR filters includes using a gradient method to optimize the Q term of each of the plurality of parametric IIR filters.

6. The method of claim 1, wherein computing parameters for a plurality of parametric IIR filters comprises computing parameters of a number of parametric filters selected from the group consisting of three and four parametric IIR filters.

7. The method of claim 1, wherein computing the center frequencies F_c comprises computing center frequencies F_c using a root finding technique applied to a polynomial produced by the LPC model.

8. The method of claim 1, further including a method for computing a 3 dB bandwidth BW and the Q for each peak comprising:

setting a counter n to 1;

comparing n to the num_peaks;

while n is less than or equal to the num_peaks, computing a gain in dB at a half bandwidth location for the nth peak;

when n is equal to 1;

finding a 3 dB bandwidth BW(1) of the first peak; and

computing the Q(1) of the first peak from the first bandwidth BW(1) and the first center frequency Fc (1);

when n is not equal to 1;

if n is less than the num_peaks, computing the 3 dB bandwidth BW(n) and the Q(n);

if n is equal to num_peak;

compute the 3 dB bandwidth BW(the num_peaks) of the last peak; and

compute the last Q factor the Q(the num_peaks) from the last bandwidth BW(the num_peaks) and the last center frequency Fc(the num_peaks).

9. The method of claim 8, wherein a method for computing the 3 dB bandwidth BW(n) and the Q(n) based on HH2_abs, the magnitude of HH2, comprises:

If the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n−

1)) is greater than 3 dB;

computing an interpolated 3 dB down points HH2_int and FF_int between the valley_loc(n−

1) and the peak_loc(n); and

computing the bandwidth BW(n) and the Q factor the Q(n) using the HH2_int and the FF_int;

if the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n−

1)) is not greater than 3 dB, and if the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n)) is greater than 3 dB;

compute the interpolated 3 dB down points HH2_int and FF_int between the valley_loc(n) and the peak_loc(n); and

compute the bandwidth BW(n) and the Q factor the Q(n) using the HH2_int and the FF_int;

if the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n−

1)) is not greater than 3 dB, and if the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n)) is not greater than 3 dB, and if the nth center frequency Fc(n) is closer to the valley_loc(n) than to the valley_loc(n−

1);

oversampling the array of frequencies FF and the HH2_abs in the region between the valley_loc(n) and Center frequency Fc(n);

finding the 3 dB downpoint between the valley_loc(n) and Center frequency Fc(n); and

computing the bandwidth BW(n) and the Q(n) at the 3 dB downpoint;

if the nth the center frequency Fc(n) is closer to the valley_loc(n−

1) than to the valley_loc(n);

oversampling the array of frequencies FF and the HH2_abs (e.g., by interpolating) in the region between the valley_loc(n−

1) and Center frequency Fc(n);

finding the 3 dB down point between the valley_loc(n−

1) and the Center frequency Fc(n); and

computing the bandwidth BW(n) and the Q(n) at the 3 dB down point.

10. A method for computing coefficients of a family of cascaded parametric IIR filters and using the filters to filter signals provided to speakers, the method comprising:

collecting unprocessed time domain room response data;

performing an FFT on the time domain room response data to obtain a frequency domain room response;

normalizing the frequency domain room response in a frequency range of interest to obtain a normalized frequency domain room response;

performing an inverse FFT on the normalized frequency domain room response to obtain a normalized time domain room response data;

representing the normalized time domain room response data using an LPC model to obtain smoothed time domain room response data;

performing an FFT on the smoothed time domain room response data to obtain smoothed frequency domain room response data;

inverting the smoothed frequency domain room response data to obtain an equalization frequency response;

computing a magnitude of the equalization frequency response;

detecting peaks and valleys of the magnitude of the equalization frequency response;

computing gains G, center frequencies F_c, bandwidths BW and bandwidth term Q factors of each of the peaks for a plurality of the parametric Infinite-duration Impulse Response (IIR) filters using a step for determining a number of peaks num_peaks, the step comprising;

computing a frequency response HH2 from a linear predication coefficient q wherein elements in the frequency response HH2 correspond to an array of frequencies FF;

given an interested frequency range between LO_FREQ and HI_FREQ in Hz, and a low bin bin_lo and a high bin, bin_hi, updating the array of frequencies FF such that the elements of the updated array of frequencies FF are frequencies in the interested frequency range;

computing HH2_abs, a magnitude of the frequency response HH2, based on the bin_lo and the bin_hi in the interested frequency range;

determining peak locations peak_loc and valley_locations valley_loc while ensuring that a first peak occurs before a first valley;

saving a number of peak locations as the num_peaks; and

determining the center frequency Fc and gain G of each peak based on the peak location peak_loc and the magnitude of the HH2_abs at the peak location;

optimizing the gains and the Q factors;

computing parametric filter coefficients from the optimized gains and the optimized Q factors;

using the parametric filter coefficients to equalize signals; and

providing the equalized signals to speakers to provide improved sound production.

11. The method of claim 10, wherein computing the center frequencies F_c comprises computing center frequencies F_c using a root finding technique.

12. The method of claim 10, further including a method for computing a 3 dB bandwidth BW and the Q for each peak comprising:

setting a counter n to 1;

comparing n to the num_peaks;

while n is less than or equal to the num_peaks, computing a gain in dB at a half bandwidth location for the nth peak;

when n is equal to 1;

finding a 3 dB bandwidth BW(1) of the first peak; and

computing the Q(1) of the first peak from the first bandwidth BW(1) and the first center frequency Fc (1);

when n is not equal to 1;

if n is less than the num_peaks, computing the 3 dB bandwidth BW(n) and the Q(n);

if n is equal to num_peak;

compute the 3 dB bandwidth BW(the num_peaks) of the last peak; and

compute the last Q factor the Q(the num_peaks) from the last bandwidth BW(the num_peaks) and the last center frequency Fc(the num_peaks).

13. The method of claim 12, wherein a method for computing the 3 dB bandwidth BW(n) and the Q(n) based on HH2_abs, the magnitude of HH2, comprises:

If the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n−

1)) is greater than 3 dB;

computing an interpolated 3 dB down points HH2_int and FF_int between the valley_loc(n−

1) and the peak_loc(n); and

computing the bandwidth BW(n) and the Q factor the Q(n) using the HH2_int and the FF_int;

if the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n−

1)) is not greater than 3 dB, and if the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n)) is greater than 3 dB;

compute the interpolated 3 dB down points HH2_int and FF_int between the valley_loc(n) and the peak_loc(n); and

compute the bandwidth BW(n) and the Q factor the Q(n) using the HH2_int and the FF_int;

if the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n−

1)) is not greater than 3 dB, and if the HH2_abs(the peak_loc(n)) minus the HH2_abs(the valley_loc(n)) is not greater than 3 dB, and if the nth center frequency Fc(n) is closer to the valley_loc(n) than to the valley_loc(n−

1);

oversampling the array of frequencies FF and the HH2_abs in the region between the valley_loc(n) and Center frequency Fc(n);

finding the 3 dB downpoint between the valley_loc(n) and Center frequency Fc(n); and

computing the bandwidth BW(n) and the Q(n) at the 3 dB downpoint;

if the nth the center frequency Fc(n) is closer to the valley_loc(n−

1) than to the valley_loc(n);

oversampling the array of frequencies FF and the HH2_abs (e.g., by interpolating) in the region between the valley_loc(n−

1) and Center frequency Fc(n);

finding the 3 dB down point between the valley_loc(n−

1) and the Center frequency Fc(n); and

computing the bandwidth BW(n) and the Q(n) at the 3 dB down point.

14. A method for computing the coefficients of a family of cascaded parametric IIR filters and using the filters to filter signals provided to speakers, the method comprising:

collecting unprocessed time domain room response data;

performing an FFT on the time domain room response data to obtain a frequency domain room response;

normalizing the frequency domain room response in a frequency range of interest to obtain a normalized frequency domain room response;

performing an inverse FFT on the normalized frequency domain room response to obtain a normalized time domain room response data;

representing the normalized time domain room response data using an LPC model to obtain smoothed time domain room response data;

performing an FFT on the smoothed time domain room response data to obtain smoothed frequency domain room response data;

computing the magnitude of the smoothed frequency domain room response;

detecting peaks and valleys of the magnitude of the smoothed frequency domain room response;

computing gains G, center frequencies F_c, bandwidths BW and bandwidth term Q factors of each of the peaks for a plurality of the parametric Infinite-duration Impulse Response (IIR) filters using a step for determining a number of peaks num_peaks, the step comprising;

computing a frequency response HH2 from a linear predication coefficient q wherein elements in the frequency response HH2 correspond to an array of frequencies FF;

given an interested frequency range between LO_FREQ and HI_FREQ in Hz, and a low bin bin_lo and a high bin, bin_hi, updating the array of frequencies FF such that the elements of the updated array of frequencies FF are frequencies in the interested frequency range;

computing HH2_abs, a magnitude of the frequency response HH2, based on the bin_lo and the bin_hi in the interested frequency range;

determining peak locations peak_loc and valley_locations valley_loc while ensuring that a first peak occurs before a first valley;

saving a number of peak locations as the num_peaks; and

determining the center frequency Fc and gain G of each peak based on the peak location peak_loc and the magnitude of the HH2_abs at the peak location;

optimizing the gains and the Q factors;

computing parametric filter coefficients from the center frequencies and the optimized gains and the optimized Q factors;

determining poles and zeros of each of the parametric IIR filters in a cascade based on the parametric filter coefficients;

computing minimum-phase zeros from the zeros of each of the parametric filters in the cascade;

reflecting each minimum-phase zero as a reflected pole and reflecting each pole as a reflected zero for each parametric filter in the cascade;

expanding each reflected zero and its complex conjugate into a real second order numerator polynomial and expanding each reflected pole and its complex conjugate into a real second order denominator polynomial for each parametric filter in the cascade;

using the parametric filters to equalize signals; and

providing the equalized signals to speakers to provide improved sound production.

15. A method for computing the coefficients of a family of cascaded parametric IIR filters and using the filters to filter signals provided to speakers, the method comprising:

collecting unprocessed time domain room response data;

performing an FFT on the time domain room response data to obtain a frequency domain room response;

normalizing the frequency domain room response in a frequency range of interest to obtain a normalized frequency domain room response;

performing an inverse FFT on the normalized frequency domain room response to obtain a normalized time domain room response data;

representing the normalized time domain room response data using an LPC model to obtain smoothed time domain room response data;

performing an FFT on the smoothed time domain room response data to obtain smoothed frequency domain room response data;

computing the magnitude of the smoothed frequency domain room response to obtain a magnitude response;

inverting the magnitude response;

detecting peaks and valleys of the inverted magnitude response;

computing gains G, center frequencies F_c, bandwidths BW and bandwidth term Q factors of each of the peaks for a plurality of the parametric Infinite-duration Impulse Response (IIR) filters using a step for determining a number of peaks num_peaks, the step comprising;

computing a frequency response HH2 from a linear predication coefficient q wherein elements in the frequency response HH2 correspond to an array of frequencies FF;

given an interested frequency range between LO_FREQ and HI_FREQ in Hz, and a low bin bin_lo and a high bin, bin_hi, updating the array of frequencies FF such that the elements of the updated array of frequencies FF are frequencies in the interested frequency range;

computing HH2_abs, a magnitude of the frequency response HH2, based on the bin_lo and the bin_hi in the interested frequency range;

determining peak locations peak_loc and valley_locations valley_loc while ensuring that a first peak occurs before a first valley;

saving a number of peak locations as the num_peaks; and

determining the center frequency Fc and gain G of each peak based on the peak location peak_loc and the magnitude of the HH2_abs at the peak location;

optimizing the gains and the Q factors;

computing parametric filter coefficients from the optimized center frequencies, the optimized gains, and the optimized Q factors;

using the parametric filters to equalize signals; and

providing the equalized signals to speakers to provide improved sound production.