Conferencing device self test
A plurality of acoustic sensors in a non-anechoic environment are calibrated with the aim of removing manufacturing tolerances and degradation over time but preserving position-dependent differences between the sensors, The sensors are excited by an acoustic stimulus which has either time-dependent characteristics or finite duration. The calibration is to be based on diffuse-field excitation only, in which indirect propagation (including single or multiple reflections) dominate over any direct-path excitation. For this purpose, the calibration process considers only a non-initial portion of sensor outputs and/or of an impulse response derived there-from. Based on these data, a frequency-dependent magnitude response function is estimated and compared with a target response function, from which a calibration function is derived.
1. A method of operating a plurality of acoustic sensors arranged in a non-anechoic environment, the method comprising the steps of:
exciting each acoustic sensor by producing, in the non-anechoic environment, an acoustic stimulus comprising a noise burst of finite duration, and concurrently sensing an output from each acoustic sensor only during a time period beginning a finite interval after the end of the acoustic stimulus;
computing a frequency-dependent magnitude response function of each acoustic sensor based on at least a non-initial portion of the output from each acoustic sensor, said computation being based on the sensed output from the acoustic sensor and a spectral composition of the noise burst; and
deriving a frequency-dependent calibration function based on a comparison of the magnitude response functions thus obtained and a target response function which is common to all acoustic sensors.
2. The method of claim 1 , further comprising evaluating an acoustic diffuseness metric based on the output from each acoustic sensor and, optionally, further based on information concerning the spatial relationship of the acoustic sensors.
3. The method of claim 1 , wherein the acoustic stimulus is produced using a plurality of sources arranged in one non-anechoic environment.
4. The method of claim 1 , further comprising:
sensing a background output from each acoustic sensor in the absence of an acoustic stimulus; and
verifying that signal powers of the outputs sensed during excitation are sufficiently greater than the signal power of each background output.
5. The method of claim 1 , further comprising:
computing signal power differences between outputs from the respective acoustic sensors; and
verifying that no signal power difference exceeds a predetermined threshold.
6. The method of claim 1 , further comprising:
computing broadband magnitude levels of the magnitude response functions of the acoustic sensors; and
computing a reference broadband magnitude level being one in the group comprising:
a) an average of the broadband magnitude levels,
b) a median of broadband magnitude levels;
and computing said target response function by normalizing a predefined target response shape by the reference broadband magnitude level.
7. The method of claim 1 , wherein the acoustic sensors are arranged in a manner ensuring spatial diversity,
the method further comprising performing a spatially directive measurement while applying the calibration function.
8. A computer readable medium storing computer-readable instructions for performing the method of claim 1 .
9. A controller connectable to a plurality of acoustic sensors arranged in a non-anechoic environment and to at least one audio source in the non-anechoic environment, the controller comprising:
a measuring section configured to excite each acoustic sensor by causing the at least one audio source to produce an acoustic stimulus comprising a noise burst of finite duration and, concurrently herewith, to capture respective outputs from the acoustic sensors by activating the acoustic sensors;
a windowing section configured to extract a non-initial time segment from the output from each acoustic sensor during a time interval beginning a finite interval after the end of the acoustic stimulus;
a magnitude response estimator configured to estimate a frequency-dependent magnitude response function of each acoustic sensor based on at least a non-initial portion of the output from each acoustic sensor supplied by the windowing section and on a spectral composition of the noise burst; and
a calibrator configured to derive, for each audio sensor, a frequency-dependent calibration function based on a comparison of the magnitude response function of the sensor and a target response function which is common to all acoustic sensors.