Voice activity detection using zero crossing detection
A first VAD system outputs a pulse stream for zero crossings in an audio signal. The pulse density of the pulse stream is evaluated to identify speech. The audio signal may have noise added to it before evaluating zero crossings. A second VAD system rectifies each audio signal sample and processes each rectified sample by updating a first statistic and evaluating the rectified sample per a first threshold condition that is a function of the first statistic. Rectified samples meeting the first threshold condition may be used to update a second statistic and the rectified sample evaluated per a second threshold condition that is a function of the second statistic. Rectified samples meeting the second threshold condition may be used to update a third statistic. The audio signal sample may be selected as speech if the second statistic is less than a downscaled third statistic.
1. An apparatus comprising:
a processing device programmed to:
receive an original audio signal;
add a high-frequency signal to the original audio signal to obtain an audio signal, the high-frequency signal having a frequency and amplitude selected to add zero crossings to portions of the original audio signal including only ambient noise without speech;
generate a pulse stream according to zero crossings detected in the audio signal;
generate a pulse density stream according to the pulse stream indicating frequency of occurrence of pulses in the pulse stream over time;
evaluating the pulse density stream with respect to a threshold condition; and
identifying speech portions of the audio signal corresponding to portions of the pulse density stream meeting the threshold condition.
2. The apparatus of claim 1 , wherein the processing device is programmed to identify the speech portions of the audio signal by identifying portions of the pulse density stream below a threshold value.
3. The apparatus of claim 1 , wherein the processing device is further programmed to:
band-pass filter the original audio signal to obtain a filtered signal; and
add the high-frequency signal to the filtered signal to obtain the audio signal.
4. The apparatus of claim 3 , wherein the high-frequency signal has a frequency equal to half a sampling frequency of the original audio signal.
5. The apparatus of claim 3 , wherein the band-pass filter has a passband between 0.3 and 20 kHz.
6. The apparatus of claim 1 , wherein the processing device is further programmed to generate the pulse density stream and evaluate the pulse density stream with respect to a threshold by:
low-pass filtering the pulse stream to obtain a filtered stream; and
identifying portions of the filtered stream having lower amplitude than the threshold condition.
7. The apparatus of claim 1 , wherein the processing device is further programmed to adjust the threshold condition according to statistical values based on the audio signal.
8. The apparatus of claim 1 , further comprising a microphone coupled to the processing device, the processing device programmed to derive the audio signal from an output of the microphone.
9. A method comprising:
receiving, by a processing device, an original audio signal;
adding, by the processing device, a high-frequency signal to the original audio signal to obtain an audio signal, the high-frequency signal having zero crossings between each pair of contiguous samples of the high-frequency signal;
generating, by the processing device, a pulse stream according to zero crossings detected in the audio signal;
generating, by the processing device, a pulse density stream according to the pulse stream indicating frequency of occurrence of pulses in the pulse stream over time;
evaluating, by the processing device, the pulse density stream with respect to a threshold condition; and
identifying, by the processing device, speech portions of the audio signal corresponding to portions of the pulse density stream meeting the threshold condition.
10. The method of claim 9 , further comprising identifying the speech portions of the audio signal by identifying portions of the pulse density stream below a threshold value.
11. The method of claim 9 , further comprising:
band-pass filtering the original audio signal with a band-pass filter to obtain a filtered signal; and
adding the high-frequency signal to the filtered signal to obtain the audio signal.
12. The method of claim 11 , wherein the band-pass filter has a passband between 0.3 and 20 kHz.
13. The method of claim 9 , wherein generating the pulse density stream and evaluating the pulse density stream with respect to the threshold condition comprises:
low-pass filtering the pulse stream to obtain a filtered stream; and
identifying portions of the filtered stream having lower amplitude than a threshold value.
14. The method of claim 9 , further comprising adjusting the threshold condition according to statistical values based on the audio signal.
15. The method of claim 9 , further comprising deriving the audio signal from an output of a microphone coupled to the processing device.