Acoustic talker localization
View Patent ↗An apparatus for locating a talker, comprising a microphone array for receiving multiple audio signals, wherein the microphone array is characterized by a predetermined beampattern, a spectral conditioner for filtering the audio signals to optimize the beampattern for talker localization, a localization estimator for calculating a localization estimate based on the filtered audio signals, an activity detector for detecting periods of speech activity, and decision logic for verifying the localization estimate during the periods of speech activity.
1. A method of locating a talker, comprising the steps of:
receiving multiple audio signals from a microphone away characterized by a predetermined beampattern;
detecting periods of speech activity in said audio signals and filtering in parallel said audio signals to optimize said predetermined beampattern for talker localization;
calculating a localization estimate based on the filtered audio signals;
discarding each localization estimate if calculated during a period where there is no speech activity; and
verifying said localization estimate during said periods of speech activity;
wherein said step of verifying includes storing a succession of localization estimates, counting occurrences of said localization estimates during said periods of speech activity and within a predetermined sample window and selecting any one of said occurrences which is repeated more than a predetermined threshold number of times within said window.
2. The method of claim 1 , wherein said step of filtering comprises subjecting said audio signals to bandpass filtering at a center frequency of 1150 Hz and a bandwidth of 300 Hz.
3. The method of claim 1 , wherein said predetermined threshold number is 12 and said predetermined sample window is 32 estimates.
4. An apparatus for locating a talker, comprising;
a microphone away for receiving multiple audio signals, said a microphone array being characterized by a predetermined beampattern;
an activity detector for detecting periods of speech activity in said audio signal;
a spectral conditioner for filtering said audio signals in parallel to said activity detector, to optimize said predetermined beampattern for talker localization; and
decision logic for discarding each localization estimate if calculated during periods where there is no speech activity and for verifying said localization estimate during said periods of speech activity;
wherein said decision logic includes a memory stack for storing up to a predetermined number of localization estimates, a counter for counting occurrences of said localization estimates during said periods of speech activity and selecting any one of said occurrences which is repeated more than a predetermined threshold number of times within said memory stack.
5. The apparatus of claim 4 , wherein said spectral conditioner comprises a bandpass filter for filtering said audio signals at a center frequency of 1150 Hz and a bandwidth of 300 Hz.
6. The apparatus of claim 4 , wherein said predetermined threshold number is 12 and said predetermined number of localization estimates within said stack is 32.