Double talk detection method based on spectral acoustic properties
View Patent ↗A method of detecting double talk condition in hands free communication devices is disclosed. In general, the method in accordance with the teachings of this invention detects double talk conditions based on inherent frequency response differences between the transducers used and acoustical effect on the spectrum of the returned echo signal. An input signal from a far-end talker and an input signal from the output from an echo canceler are received by the detector. K spectral subbands are created for each input signal. From this K subbands q subbands are selected based on inherent frequency differences between the far-end transducer and a near-end transducer. The spectral echo residual power is estimated at each subband. The estimated spectral echo power and the output signal from the echo canceler for a selected subband are compared to a predetermined threshold. Based on this comparison, it is determined whether double talk conditions exist based on the comparison.
1. A method of detecting double talk conditions in a hands free communication system, the method comprising:
receiving an input signal from a far-end transducer and an input signal from the output from an echo canceller;
converting the input signals to a frequency domain using N-point FFT blocks;
creating K spectral subbands from output of FFT blocks;
selecting q spectral subbands for each converted input signal based on inherent frequency differences between the far-end transducer and a near-end transducer; and
determining whether double talk conditions exist based on a comparison of the estimated echo power from far-end transducer to average power output of echo canceller for selected q subbands.
2. The method of claim 1 wherein the step of determining double talk comprises:
estimating the spectral echo power for selected subbands for the input signal from the far-end transducer;
estimating average power output for selected subband for the input signal from the output of the echo canceller; and
comparing the spectral echo power and the average power output to a predetermined threshold.
3. The method of claim 2 , further comprising estimating the echo loss coefficients (ELC) for each spectral subband, wherein the spectral echo power at each subband is estimated based on the estimated ELCs.
4. The method of claim 1 , wherein the ELCs are estimated using a recursive method.
5. The method of claim 4 , wherein K is less than N.
6. The method of claim 1 , further comprising the step of estimating average power output of each subband prior to the step of estimating the ELCs.
7. The method of claim 6 , wherein the step of estimating average power output of each subband is done using a recursive method.
8. The method of claim 1 , further comprising the step of estimating average noise power for the input signal from the output from the echo canceller in each subband prior to the comparing step.
9. The method of claim 8 , wherein the step of estimating average noise power for the input signal from the output from the echo canceller in each subband is done using a recursive method.
10. The method of claim 9 , further comprising the step of calculating an average estimate of background noise and subtracting this from the input signal from the output from the echo canceller to reduce sensitivity.
11. The method of claim 1 , further comprising the step of automatically finding subbands where echo power is less than a near-end signal power.
12. A double talk detector, comprising: a circuit to detect inherent frequency differences between a far-end transducer and a near-end transducer in a hands free communication system; a circuit to detect acoustical effect on the spectrum of the returned echo signal; and a circuit to compare the inherent frequency differences and acoustical effect to detect double talk conditions.
13. The detector of claim 12 , comprising:
an input for receiving an input signal from a far-end transducer and an input signal from the output from an echo canceller;
a FFT filter for converting the input signals to a frequency domain using N-point complex FFT blocks;
a sub bander unit for creating K spectral subbands for each input signal;
a circuit for selecting q subbands based on inherent spectral acoustic differences between the far-end transducer and a near-end transducer;
an estimator for estimating the spectral echo power for selected subbands for the input signal from the far-end transducer;
an estimator for estimating average power output for selected subbands for the input signal from the output of the echo canceller;
a comparator for comparing the spectral echo power and the average power output to a predetermined threshold; and
a circuit for determining whether double talk conditions exist based on the comparison.
14. The detector of claim 13 , wherein the circuit for estimating the spectral echo residual power at each subband comprises a circuit for estimating the echo loss coefficients (ELC) for each spectral subband, wherein the spectral echo residual power at each subband is estimated based on the estimated ELCs.
15. The detector of claim 14 , further comprising a circuit for estimating average power output of each subband prior to estimating the ELCs.
16. The detector of claim 15 , wherein the method of estimating average power output of each subband is a recursive circuit.
17. The detector of claim 16 , further comprising a noise estimation circuit for estimating average noise power for the input signal from the output from the echo canceller in each subband prior to comparing.
18. The detector of claim 17 , wherein the noise estimation circuit is a recursive circuit.
19. The detector of claim 18 , further comprising a circuit for calculating an average estimate of background noise and subtracting this from the input signal from the output from the echo canceller to reduce sensitivity.
20. The detector of claim 19 , further comprising a circuit for automatically finding frequency bands where echo power is less than a near-end signal power.
21. The detector of claim 14 , wherein the circuit for estimating the echo loss coefficients (ELC) for each spectral subband and the circuit for estimating the spectral echo residual power at each subband based on each ELC are combined into a single circuit.
22. A method of detecting double talk conditions in a hands free communication system, the method comprising: detecting inherent frequency differences between a far-end transducer and a near-end transducer in a hands free communication system; detecting acoustical effect on the spectrum of the returned echo signal; and comparing the inherent frequency differences and acoustical effect to detect double talk conditions.