Methods and apparatus to determine locations of audience members
View Patent ↗A disclosed example method to determine a location of an audience member involves generating a correlation analysis result based on correlating first audio samples from a stationary audio detector with second audio samples from a portable audio detector carried by the audience member and determining via a neural network the location of the audience member based on the correlation analysis result.
1. A method to determine a location of an audience member, the method comprising:
generating a correlation analysis result based on correlating first audio samples from a stationary audio detector with second audio samples from a portable audio detector carried by the audience member;
determining via a neural network the location of the audience member based on the correlation analysis result; and
crediting media presented via a media delivery device with an exposure credit when the audience member was located within a threshold distance of the media delivery device as indicated by the location of the audience member.
2. A method as defined in claim 1 , wherein generating the correlation analysis result comprises:
determining a first sum of squares of each audio sample in the first audio samples;
multiplying the first sum of squares by a quantity value representing a total audio sample quantity of the first and second audio samples to determine a first product;
determining a square root of the first product;
determining a second sum of squares of each audio sample in the second audio samples;
multiplying the second sum of squares by the quantity value to determine a second product;
determining a square of a third sum of the second audio samples;
determining a second square root of a difference between the second product and the square of the third sum;
multiplying the first square root by the second square root to determine a third product;
determining a fourth sum of fourth products between corresponding audio samples of the first and second audio samples;
multiplying the fourth sum by the quantity value to determine a fifth product; and
determining a correlation coefficient by dividing the fifth product by the third product.
3. A method as defined in claim 2 , wherein the correlation analysis result is determined when an average value of the first audio samples is zero and an average value of the second audio samples is not zero.
4. A method as defined in claim 1 , wherein the first audio samples are a first sub-sample set of first total audio samples collected by the stationary audio detector and the second audio samples are a second sub-sample set of second total audio samples collected by the portable audio detector.
5. A method as defined in claim 1 , wherein determining the location of the audience member is based on a first maximum-amplitude audio sample from the first audio samples and a second maximum-amplitude audio sample from the second audio samples.
6. A method as defined in claim 1 , further comprising:
determining a first frequency-domain representation of the first audio samples and a second frequency-domain representation of the second audio samples before generating the correlation analysis result; and
generating the correlation analysis result by correlating the first frequency-domain representation of the first audio samples with the second frequency-domain representation of the second audio samples.
7. A method as defined in claim 6 , wherein generating the correlation analysis result comprises:
determining a first sum of complex conjugate pairs of the first frequency-domain representation;
determining a second sum of complex conjugate pairs of the second frequency-domain representation;
multiplying the first sum of complex conjugate pairs by the second sum of complex conjugate pairs to determine a first product;
determining a square root of the first product;
multiplying complex numbers of the first frequency-domain representation by corresponding conjugates of the second frequency-domain representation to determine second products;
determining a third sum of the second products; and
determining a correlation coefficient associated with the first and second audio samples by dividing the third sum of the second products by the square root of the first product.
8. A method as defined in claim 7 , wherein the correlation analysis result is determined when an average value of the first audio samples is zero and an average value of the second audio samples is zero.
9. A method as defined in claim 6 , wherein generating the correlation analysis result comprises:
determining a first sum of complex conjugate pairs of the first frequency-domain representation;
multiplying the first sum of complex conjugate pairs by a quantity value representing a total quantity of audio samples of the first and second audio samples to determine a first product;
determining a first square root of the first product;
determining a second sum of complex conjugate pairs of the second frequency-domain representation;
multiplying the second sum of complex conjugate pairs by the quantity value to determine a second product;
determining a third sum of complex numbers of the second frequency-domain representation;
determining a conjugate of the third sum of complex numbers;
multiplying the third sum of complex numbers by the conjugate to generate a third product;
determining a second square root of a difference between the second product and the third product;
multiplying the first square root by the second square root to determine a third product;
multiplying complex numbers of the first frequency-domain representation by corresponding conjugates of the second frequency-domain representation to determine fourth products;
determining a fourth sum of the fourth products;
multiplying the fourth sum by the quantity value to determine a fifth product; and
determining a correlation coefficient by dividing the fifth product by the third product.
10. A method as defined in claim 9 , wherein the correlation analysis result is determined when an average value of the first audio samples is zero and an average value of the second audio samples is not zero.
11. A method as defined in claim 6 , wherein generating the correlation analysis result comprises:
determining a first sum of reference complex conjugate pairs;
multiplying the first sum by a quantity value representing a total quantity of audio samples of the first and second audio samples to determine a first product;
determining a second sum of reference complex numbers;
determining a conjugate of a third sum of complex numbers of the first frequency-domain representation;
multiplying the second sum by the conjugate of the third sum to determine a second product;
determining a first square root of a difference between the first and second products;
determining a fourth sum of complex conjugate pairs of the second frequency-domain representation;
multiplying the fourth sum by the quantity value to determine a third product;
determining a fifth sum of complex numbers of the second frequency-domain representation;
determining a conjugate of a sixth sum of complex numbers of the second frequency-domain representation;
multiplying the fifth and sixth sums to determine a fourth product;
determining a second square root of a difference between the third and fourth products;
multiplying the first and second square roots to determine a fifth product;
multiplying complex numbers of the first frequency-domain representation by corresponding conjugates of the second frequency-domain representation to determine sixth products;
determining a seventh sum of the sixth products;
multiplying the seventh sum by the quantity value;
determining an eighth sum of conjugates of the second frequency-domain representation;
multiplying the second sum by the eighth sum to determine a seventh product; and
dividing a difference between the eighth product and the seventh product by the fifth product.
12. A method to determine a location of an audience member, the method comprising:
generating a correlation analysis result based on correlating first audio samples from a stationary audio detector with second audio samples from a portable audio detector carried by the audience member; and
determining via a neural network the location of the audience member based on the correlation analysis result,
wherein generating the correlation analysis result comprises:
determining a first sum of squares of each audio sample in the first audio samples;
determining a second sum of squares of each audio sample in the second audio samples;
multiplying the first sum of squares by the second sum of squares to determine a first product;
determining a square root of the first product;
determining a third sum of second products between corresponding audio samples of the first and second audio samples; and
determining a correlation coefficient associated with the first and second audio samples by dividing the third sum of the second products by the square root of the first product.
13. A method as defined in claim 12 , wherein generating the correlation analysis result is performed when an average value of the first audio samples is zero and an average value of the second audio samples is zero.
14. An apparatus to determine a location of an audience member, the apparatus comprising:
a correlator structured to generate a correlation analysis result based on correlating first audio samples from a stationary audio detector with second audio samples from a portable audio detector carried by the audience member;
a neural network structured to determine the location of the audience member based on the correlation analysis result; and
a processor to credit media presented via a media delivery device with an exposure credit when the audience member was located within a threshold distance of the media delivery device as indicated by the location of the audience member.
15. An apparatus as defined in claim 14 , wherein the correlator is to generate the correlation analysis result by:
determining a first sum of squares of each audio sample in the first audio samples;
multiplying the first sum of squares by a quantity value representing a total audio sample quantity of the first and second audio samples to determine a first product;
determining a square root of the first product;
determining a second sum of squares of each audio sample in the second audio samples;
multiplying the second sum of squares by the quantity value to determine a second product;
determining a square of a third sum of the second audio samples;
determining a second square root of a difference between the second product and the square of the third sum;
multiplying the first square root by the second square root to determine a third product;
determining a fourth sum of fourth products between corresponding audio samples of the first and second audio samples;
multiplying the fourth sum by the quantity value to determine a fifth product; and
determining a correlation coefficient by dividing the fifth product by the third product.
16. An apparatus to determine a location of an audience member, the apparatus comprising:
a correlator structured to generate a correlation analysis result based on correlating first audio samples from a stationary audio detector with second audio samples from a portable audio detector carried by the audience member; and
a neural network structured to determine the location of the audience member based on the correlation analysis result,
wherein the correlator is to generate the correlation analysis result by:
determining a first sum of squares of each audio sample in the first audio samples;
determining a second sum of squares of each audio sample in the second audio samples;
multiplying the first sum of squares by the second sum of squares to determine a first product;
determining a square root of the first product;
determining a third sum of second products between corresponding audio samples of the first and second audio samples; and
determining a correlation coefficient associated with the first and second audio samples by dividing the third sum of the second products by the square root of the first product.
17. A tangible machine accessible storage medium comprising instructions that, when executed, cause a machine to at least:
generate a correlation analysis result based on correlating first audio samples from a stationary audio detector with second audio samples from a portable audio detector carried by the audience member;
determine via a neural network the location of the audience member based on the correlation analysis result; and
credit media presented via a media delivery device with an exposure credit when the audience member was located within a threshold distance of the media delivery device as indicated by the location of the audience member.
18. The tangible machine accessible storage medium of claim 17 , wherein the first audio samples are a first sub-sample set of first total audio samples collected by the stationary audio detector and the second audio samples are a second sub-sample set of second total audio samples collected by the portable audio detector.
19. The tangible machine accessible storage medium of claim 17 , wherein the instructions further cause the machine to determine the location of the audience member based on a first maximum-amplitude audio sample from the first audio samples and a second maximum-amplitude audio sample from the second audio samples.
20. The tangible machine accessible storage medium of claim 17 , wherein the instructions further cause the machine to:
determine a first frequency-domain representation of the first audio samples and a second frequency-domain representation of the second audio samples before generating the correlation analysis result, the correlation analysis result being generated based on correlating the first frequency-domain representation of the first audio samples with the second frequency-domain representation of the second audio samples.