Adaptive equalizer preprocessor for mobile telephone speech coder to modify nonideal frequency response of acoustic transducer

• US 5,915,235 A
• Filed: 10/17/1997
• Issued: 06/22/1999
• Est. Priority Date: 04/28/1995
• Status: Expired due to Term

• Current Assignee: Andrew P. Dejaco, John Miller
• Original Assignee: Andrew P. Dejaco, John Miller
• Inventors: Miller, John A., DeJaco, Andrew P.
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): Smits, Talivaldis Ivars
• Application Number:
  US08/953,102
• Time in Patent Office: 613 Days
• Field of Search: 704/219, 704/228, 704/234, 704/203, 375/232, 342/90
• US Class Current: 704/234
• CPC Class Codes: G10L 21/0364 for improving intelligibility

Abstract

The present invention teaches an equalizer preprocessor for a mobile telephone speech coder that adapts to the characteristics of its input transducer. The equalizer determines the frequency response of the input transducer by measuring the long term characteristics of the input signal and estimating the spectral envelope of that signal. The equalizer then adapts so that the output signal has a spectral response closer to a perceptually ideal response in accordance with the calculated spectral envelope. In a first embodiment of the present invention, the adaptive equalizer is implemented using digital filtering techniques. The equalizer determines a set of long term autocorrelation coefficient values and from these values generates a set of filter taps which serve to whiten or flatten the spectral response of the input signal. This whitened signal is then passed through a target filter which impresses upon the whitened signal the target spectral response. In an alternative embodiment, the equalizer is realized by using a bank of variable gain control elements to adjust the energy of subbands of the input signal. A subband filter bank divides the input signal into frequency subbands. Each of the subbands is then provided to a corresponding variable gain stage element and the energy of the subband is amplified or reduced depending upon a corresponding subband gain signals. The subband gain signals are determined in accordance with the long term subband energy and a target subband energy.

Claims

1. In a mobile telephone, an apparatus for encoding a speech signal, comprising:

(A) an acoustic to electrical transducer that receives the speech signal and converts the speech signal to an electrical signal, the acoustic to electrical transducer having a frequency response that is different from an ideal frequency response;

(B) an adaptive equalizer that receives an input signal representative of the electrical signal from the acoustic to electrical transducer, the adaptive equalizer including;

first subband filter means for receiving said input signal and for bandpass filtering said input signal in accordance with a first bandpass frequency format to output a first subband signal;

first variable gain means for receiving said first subband signal and for amplifying said first subband signal in accordance with a first subband target gain to output a first gain adjusted subband signal;

at least one additional subband filter means for receiving said input signal and for bandpass filtering said input signal in accordance with at least one additional bandpass frequency format to output at least one additional subband signal;

at least one additional variable gain means for receiving said at least one additional subband gain signal and for amplifying said at least one additional subband gain signal in accordance with at least one additional target subband gain to output at least one additional gain adjusted subband signal; and

summing means for receiving said first gain adjusted subband signal and said at least one additional gain adjusted subband signal and for summing said gain adjusted signals to provide an equalized signal having a spectrum that is closer to said ideal frequency response;

(C) a speech encoder that encodes the equalized signal from the adaptive equalizer.

2. The apparatus of claim 1 further comprising:

first subband gain calculator means for receiving said first subband signal and computing a long term subband energy in accordance with said first subband signal and for computing said target subband gain value in accordance with said long term subband energy and a first target subband energy; and

at least one additional subband gain calculator means for receiving said at least one additional subband signal and computing at least one additional long term subband energy in accordance with said at least one additional subband signal and for computing at least one additional target subband gain value in accordance with said long term subband energy and a at least one additional target subband energy.

3. The apparatus of claim 2, further comprising:

subband energy calculator means for receiving said subband signals, measuring the energy of said subband signals, and determining a maximum energy of said subband signals,

wherein said subband gain calculator means compute said target subband gain values further in accordance with said maximum energy.

4. The apparatus of claim 1, further comprising:

gain means for determining a gain factor for adjusting the energy of said equalized signal to generate a gain adjusted output signal which has the same long term energy level as said input signal.

5. The apparatus of claim 4 wherein said gain factor is based on a ratio of the energy of said input signal and the energy of said equalized signal.

6. In a mobile telephone, an apparatus for encoding a speech signal, comprising:

(A) an acoustic to electrical transducer that receives the speech signal and converts the speech signal to an electrical signal, the acoustic to electrical transducer having a frequency response that is different from an ideal frequency response;

(B) an adaptive equalizer that receives an input signal representative of the electrical signal from the acoustic to electrical transducer, the adaptive equalizer including;

adaptive whitening filter means for receiving said input signal and for filtering said input signal in order to flatten a long term spectral response of said input signal to provide a whitened signal; and

target filter means for receiving said whitened signal and for filtering said whitened signal in accordance with a target spectral response to provide a target filtered signal, wherein said target spectral response is for impressing a spectrum that is closer to said ideal frequency response upon said whitened signal;

(C) a speech encoder that encodes the target filtered signal from the adaptive equalizer.

7. The adaptive equalizer of claim 6, further comprising:

input spectral response means for receiving said input signal and for computing said long term spectral response in accordance with said input signal.

8. The adaptive equalizer of claim 7, further comprising:

filter tap calculator means for generating filter coefficient values for said adaptive whitening filter responsive to said input signal.

9. The adaptive equalizer of claim 8 wherein said filter tap calculator means generates said filter coefficient values in accordance with a linear prediction coding (LPC) format.

10. The adaptive equalizer of claim 9 wherein said filter tap calculator means generates said filter coefficient values in accordance with long term autocorrelation coefficients.

11. The adaptive equalizer of claim 7, further comprising:

gain calculator means for determining a gain factor for adjusting the energy of said target filtered signal so that an output equalized signal has the same long term energy level as said input signal; and

variable gain stage means for imposing said gain factor upon said target filtered signal to provide said output equalized signal.

12. The adaptive equalizer of claim 11 wherein said gain factor is based on a ratio of the long term energy of said input signal and the long term energy of said target filtered signal.

13. In a mobile telephone, a method for encoding a speech signal using adaptive equalization, comprising the steps of:

providing the speech signal to an acoustic to electrical transducer that converts the speech signal to an electrical signal, the acoustic to electrical transducer having a frequency response that is different from an ideal frequency response;

filtering an input signal representative of the electrical signal from the acoustic to electrical transducer in order to flatten a long term spectral response of said input signal to provide a whitened signal;

filtering said whitened signal in accordance with a target spectral response to provide a target filtered signal, wherein said target spectral response impresses a spectrum that is closer to said ideal frequency response upon said whitened signal;

(C) encoding, with a speech encoder, the target filtered signal from the adaptive equalizer.

14. The method of claim 13, further comprising the step of:

determining said long term spectral response in accordance with said input signal.

15. The method of claim 14, further comprising the step of:

generating filter coefficient values responsive to said input signal for filtering said input signal to provide said whitened signal.

16. The method of claim 15 wherein said step of generating filter coefficient values generates said filter coefficient values in accordance with a linear prediction coding (LPC) format.

17. The method of claim 15 wherein said step of generating filter coefficient values generates said filter coefficient values in accordance with long term autocorrelation coefficients.

18. The method of claim 14, further comprising the steps of:

determining a gain factor for adjusting the energy of said target filtered signal so that an output equalized signal has the same long term energy level as said input signal; and

adjusting the gain of said target filtered signal based on said gain factor to provide said output equalized signal.

19. The method of claim 18 wherein said step of determining a gain factor determines the gain factor to be based on a ratio of the long term energy of said input signal and the long term energy of said target filtered signal.

20. In a mobile telephone, an apparatus for encoding a speech signal, comprising:

(A) an acoustic to electrical transducer that receives the speech signal and converts the speech signal to an electrical signal, the acoustic to electrical transducer having a frequency response that is different from an ideal frequency response;

(B) an adaptive equalizer that receives an input signal representative of the electrical signal from the acoustic to electrical transducer, the adaptive equalizer including;

an adaptive whitening filter for receiving said input signal and for filtering said input signal in order to flatten a long term spectral response of said input signal to provide a whitened signal; and

a target filter for receiving said whitened signal and for filtering said whitened signal in accordance with a target spectral response to provide a target filtered signal, wherein said target spectral response is for impressing a spectrum that is closer to said ideal frequency response upon said whitened signal;

(C) a speech encoder that encodes the target filtered signal from the adaptive equalizer.

21. The adaptive equalizer of claim 20, further comprising:

an input spectral response element for receiving said input signal and for computing said long term spectral response in accordance with said input signal.