Methods and apparatus to classify audio
View Patent ↗Methods, apparatus, systems and articles of manufacture are disclosed herein to classify audio. An example method includes determining a first count of volume events having a first period of time associated with the audio. The example method also includes determining a second count of volume events having a second period of time associated with the audio. The example method also includes classifying the audio based on the first count and the second count.
1. A method to classify audio, the method comprising:
associating, by executing an instruction with a processor, blocks of audio sensed from a media presentation
device with volume events, a first audio block being associated with a first volume event when (1) the first audio block has a first volume value satisfying a volume threshold and (2) a ratio of the first volume value and a second volume value of a second audio block associated with a second volume event satisfies an event threshold;
determining, by executing an instruction with the processor, a first count of volume events having respective elapsed times corresponding to a first range, a respective elapsed time of the first volume event corresponding to a first period of time between the first audio block and the second audio block;
determining, by executing an instruction with the processor, a second count of volume events having respective elapsed times corresponding to a second range; and
classifying, by executing an instruction with the processor, the audio sensed from the media presentation device based on the first count of volume and the second count of volume.
2. The method as defined in claim 1 , further including:
accumulating samples from the audio to form the first audio block; and
summing absolute values of the samples of the first audio block to determine the first volume value.
3. The method as defined in claim 1 , wherein the first volume value is determined subsequent to the second volume value.
4. The method as defined in claim 1 , further including, when the first volume value satisfies the volume threshold, retrieving the second volume value from a buffer to determine whether the first audio block is associated with the first volume event.
5. The method as defined in claim 4 , wherein the first volume value is stored in the buffer when the first volume value satisfies the volume threshold.
6. An apparatus to classify audio, the apparatus comprising:
a volume comparator to associate blocks of audio sensed from a media presentation device with volume events, the volume comparator to associate a first audio block with a first volume event when (1) the first audio block has a first volume value that satisfies a volume threshold and (2) a ratio of the first volume value and a second volume value of a second audio block associated with a second volume event satisfies an event threshold;
an adder to:
determine a first count of volume events that have respective elapsed times that correspond to a first range, a respective elapsed time of the first volume event to correspond to a first period of time between the first audio block and the second audio block; and
determine a second count of volume events that have respective elapsed times that correspond to a second range; and
a classifier to classify the audio sensed from the media presentation device based on the first count of volume and the second count.
7. The apparatus as defined in claim 6 , further including a volume calculator to:
accumulate samples of the audio to form the first audio block; and
sum absolute values of the samples of the first audio block to determine the first volume value.
8. A tangible computer readable storage medium comprising instructions that, when executed, cause a processor to at least:
associate blocks of audio sensed from a media presentation device with volume events, the processor to associate a first audio block with a first volume event when (1) the first audio block has a first volume value that satisfies a volume threshold and (2) a ratio of the first volume value and a second volume value of a second audio block associated with a second volume event satisfies an event threshold;
determine a first count of volume events that have respective elapsed times that correspond to a first range, a respective elapsed time of the first volume event to correspond to a first period of time between the first audio block and the second audio block;
determine a second count of volume events that have respective elapsed times that correspond to a second range; and
classify the audio sensed from the media presentation device based on the first count and the second count.
9. The tangible computer readable storage medium as defined in claim 8 , wherein the instructions, when executed, further cause the processor to:
accumulate samples from the audio to form the first audio block; and
sum absolute values of the samples of the first audio block to determine the first volume value.
10. The method as defined in claim 1 , further including comparing the first volume value to a lowest volume value to reduce false volume event detections.
11. The method as defined in claim 1 , wherein the respective elapsed time of the first volume event corresponds to a number of audio blocks between the first audio block and the second audio block.
12. The method as defined in claim 1 , wherein the classifying of the audio includes:
identifying the audio as at least one of speech or not speech; and
in response to identifying the audio as speech, determining the audio is collected from a television program presentation.
13. The method as defined in claim 12 , further including determining the audio is collected from a video game presentation when the audio is identified as not speech.
14. The method as defined in claim 1 , wherein the classifying of the audio includes mapping the first count of volume and the second count of volume to a classification in a multi-dimensional lookup table.
15. The method as defined in claim 1 , wherein the classifying of the audio includes calculating a probability of the audio corresponding to a television program presentation.
16. The apparatus as defined in claim 6 , wherein the classifier is further to:
identify the audio as speech or not speech; and
in response to identification of the audio as speech, determine the audio is collected from a television program presentation.
17. The apparatus as defined in claim 6 , wherein the classifier is further to map the first count and the second count to a classification in a multi-dimensional lookup table.
18. The apparatus as defined in claim 6 , wherein the classifier is further to calculate a probability of the audio corresponding to a television program presentation.
19. The tangible computer readable storage medium as defined in claim 8 , wherein the instructions further cause the processor to:
identify the audio as speech or not speech; and
in response to identification of the audio as speech, determine the audio is collected from a television program presentation.
20. The tangible computer readable storage medium as defined in claim 8 , wherein the instructions, when executed, further cause the processor to map the first count and the second count to a classification in a multi-dimensional lookup table.
21. The tangible computer readable storage medium as defined in claim 8 , wherein the instructions further cause the processor to calculate a probability of the audio corresponding to a television program presentation.