Method and apparatus for performing echo suppression
View Patent ↗An echo suppression technique is provided in which speech sample energies are collected and accumulated over predetermined time frames. From this, a peak aggregate value and time delay are noted. This peak is generated by mathematical correlation between the historical output speech and the input speech. A moving average is applied to the results of the mathematical correlation. Once the time delay and gain are determined, the algorithm of the present invention uses this information to determine if an input sample is echo or a valid input, and if it is echo, the echo is suppressed.
1. A method of performing echo suppression in a telecommunications system, the method comprising:
(A) calculating at least one value representing energy in a plurality of pulse code modulated (PCM) samples of an input speech signal;
(B) building a synthetic echo envelope from said values;
(C) aggregating said values for the at least one sample over a period of time to form a frame of an aggregate energy value for that period of time;
(D) solving a plurality of normal equations for said aggregated values, the plurality of normal equations having a plurality of results;
(E) examining the results to determine a peak aggregate result, the peak aggregate indicating a time delay and a gain of an echo path; and
(F) evaluating each incoming PCM sample against a corresponding output energy result obtained at a determined time delay, and if an input speech energy is determined to be less than a historical output energy scaled by a determined gain, then suppressing echo from the input speech signal.
2. The method of performing echo suppression as defined in claim 1 including the further step of smoothing the results of the normal equations by applying a moving average to correlations and energies over each frame across the time dimension.
3. The method of performing echo suppression as defined in claim 1 including the further step of determining said time delay by measuring the time elapsed between the beginning of measurements and the reaching of the peak aggregate result.
4. The method of performing echo suppression as defined in claim 1 including the further step of employing a voice activity detector to verify that voice information is on the line and if so, then performing steps A through F and suppressing any echo that is determined to exist.
5. An apparatus for performing echo suppression techniques in a telecommunications system, the apparatus comprising:
(A) a receiver that receives a plurality of pulse code modulated (PCM) samples of voice information from a user coupled with the system;
(B) an energy accumulator coupled to said receiver that calculates at least one energy value for input speech signals and aggregates the energy values over a predetermined time period to create a synthetic echo envelope from said samples;
(C) digital signal processing circuitry coupled with said receiver and said energy accumulator adapted to:
(i) evaluate said energy values to find a peak aggregate value and a time lag; and
(ii) check each PCM sample against said peak aggregate value and time lag to determine whether said speech samples contain an echo; and
(D) an echo suppressor responsive to said digital signal processing circuitry for suppressing the echo that is determined to exist in an incoming speech sample.
6. The apparatus for performing echo suppression techniques as defined in claim 5 further comprising
a voice activity detector coupled with said receiver that determines whether incoming samples contain speech, and if so, said echo suppression techniques are performed.
7. A system for performing echo suppression techniques, the system comprising
(A) a receiver that receives pulse code modulated (PCM) samples of voice information from a user coupled with the system;
(B) an energy accumulator coupled to said receiver that calculates energy values of input speech signals and aggregates these energy values over a predetermined time period to create a synthetic echo envelope from said samples;
(C) digital signal processing circuitry coupled with said receiver and said energy accumulator that is adapted to:
(i) evaluate said energy values to find a peak aggregate value and a time lag; and
(ii) check each incoming PCM sample against said peak aggregate value and time lag to determine whether said speech samples contain echo; and
(D) an echo suppressor responsive to digital signal processing for suppressing echo that is determined to exist in an incoming speech sample.
8. The system for performing echo suppression techniques as defined in claim 7 further comprising a voice activity detector coupled with said receiver that determines whether incoming samples contain speech, and if so, said echo suppression techniques are performed.