Methods and apparatus to perform signature matching using noise cancellation models to achieve consensus
Examples are disclosed to perform signature matching using noise cancellation models to achieve consensus. Example apparatus disclosed herein include a signature matcher to compare a first stream of monitored media signatures to streams of reference signatures representative of corresponding reference media to determine a first signature match, and compare a second stream of monitored media signatures to the streams of reference signatures to determine a second signature match; a match selector to use at least one the first signature match or the second signature match to identify a first one of the reference media corresponding to the monitored media data; and a creditor interface to output identification data for the first one of the reference media identified with the at least one the first signature match or the second signature match, the identification data to be used to credit a media exposure corresponding to the monitored media.
1. A device comprising:
at least one processor; and
memory having stored therein machine readable instructions that, when executed by the at least one processor, cause the device to:
process unmodified audio collected during a time interval via at least one microphone of a meter with a first noise reduction model to obtain a first version of audio;
process the unmodified audio with a second noise reduction model to generate a modified version of the audio, the second noise reduction model different from the first noise reduction model;
access, from the meter associated with a media device, (i) a first sequence of monitored media signatures determined by the meter based on the first version of the audio captured by the meter during the time interval, and (ii) a second sequence of monitored media signatures determined by the meter based on the modified version of the audio generated by the meter, the modified version of the audio based on the first version of the audio captured by the meter during the time interval;
compare the first sequence of monitored media signatures with a plurality of sequences of reference media signatures to determine a first match result;
compare the second sequence of monitored media signatures with the plurality of sequences of reference media signatures to determine a second match result; and
credit, based on the first match result and the second match result, media exposure associated with the media device.
2. The device of claim 1 , wherein the first version of the audio captured by the meter during the time interval is ambient audio, and wherein the modified version of the audio generated by the meter is generated by application of the second noise reduction model to the unmodified audio.
3. The device of claim 1 , wherein the first match result is associated with a first media source identifier, and the second match result is associated with a second media source identifier, and wherein the machine readable instructions further cause, when executed by the at least one processor, the device to identify, based on the first media source identifier and the second media source identifier, a source of media presented by the media device during the time interval.
4. The device of claim 3 , wherein the first match result is associated with a first match strength, and the second match result is associated with a second match strength, and wherein the machine readable instructions further cause, when executed by the at least one processor, the device to identify the source of media based on the first media source identifier, the second media source identifier, the first match strength, and the second match strength.
5. The device of claim 4 , wherein the first match strength is based on a first number of individual signature matches associated with the first match result, and the second match strength is based on a second number of individual signature matches associated with the second match result.
6. The device of claim 3 , wherein the first media source identifier is associated with a first sequence of reference media signatures determined to match the first sequence of monitored media signatures, and wherein the second media source identifier is associated with a second sequence of reference media signatures determined to match the second sequence of monitored media signatures.
7. A method comprising:
processing unmodified audio collected during a time interval via at least one microphone of a meter with a first noise reduction model to obtain a first version of audio;
processing the unmodified audio with a second noise reduction model to generate a modified version of the audio, the second noise reduction model different from the first noise reduction model;
accessing, from the meter associated with a media device, (i) a first sequence of monitored media signatures determined by the meter based on the first version of the audio captured by the meter during the time interval, and (ii) a second sequence of monitored media signatures determined by the meter based on the modified version of the audio generated by the meter, the modified version of the audio based on the first version of the audio captured by the meter during the time interval;
comparing, via execution of instructions with at least one processor, the first sequence of monitored media signatures with a plurality of sequences of reference media signatures to determine a first match result;
comparing, via execution of instructions with the at least one processor, the second sequence of monitored media signatures with the plurality of sequences of reference media signatures to determine a second match result; and
crediting, based on the first match result and the second match result, media exposure associated with the media device.
8. The method of claim 7 , wherein the first version of the audio captured by the meter during the time interval is ambient audio, and wherein the modified version of the audio generated by the meter is generated by application of the second noise reduction model to the unmodified audio.
9. The method of claim 7 , wherein the first match result is associated with a first media source identifier, and the second match result is associated with a second media source identifier, and wherein the method further includes identifying, based on the first media source identifier and the second media source identifier, a source of media presented by the media device during the time interval.
10. The method of claim 9 , wherein the first match result is associated with a first match strength, and the second match result is associated with a second match strength, and wherein the method further includes identifying the source of media based on the first media source identifier, the second media source identifier, the first match strength, and the second match strength.
11. The method of claim 10 , further including:
determining the first match strength based on a first number of individual signature matches associated with the first match result; and
determining the second match strength based on a second number of individual signature matches associated with the second match result.
12. The method of claim 9 , wherein the first media source identifier is associated with a first sequence of reference media signatures determined to match the first sequence of monitored media signatures, and wherein the second media source identifier is associated with a second sequence of reference media signatures determined to match the second sequence of monitored media signatures.
13. A meter comprising:
a network interface;
at least one microphone;
at least one processor;
memory having stored thereon machine readable instructions that, when executed by the at least one processor, cause the meter to:
process unmodified audio collected during a time interval via at least one microphone with a first noise reduction model to obtain a first version of audio;
generate a first sequence of monitored media signatures based on the first version of the audio collected via the at least one microphone during the time interval;
process the unmodified audio with a second noise reduction model to generate a modified version of the audio, the second noise reduction model different from the first noise reduction model;
generate the modified version of the audio collected during the time interval, wherein the modified version differs from the first version;
generate a second sequence of monitored media signatures based on the modified version; and
report, via the network interface, the first sequence of monitored media signatures and the second sequence of monitored media signatures.
14. The meter of claim 13 , wherein the first version of the audio is the unmodified audio collected via the at least one microphone during the time interval, and wherein the modified version is generated by applying the second noise reduction model to the unmodified audio.
15. The meter of claim 13 , wherein the machine readable instructions further cause, when executed by the at least one processor, the meter to:
process the unmodified audio collected during the time interval via the at least one microphone with a first combination of noise reduction models to obtain the first version of the audio; and
process the unmodified audio with a second combination of noise reduction models to generate the modified version of the audio, the second combination different from the first combination.
16. The meter of claim 13 , wherein the machine readable instructions further cause, when executed by the at least one processor, the meter to:
associate a timestamp with the first sequence of monitored media signatures; and
associate the timestamp with the second sequence of monitored media signatures.
17. A method comprising:
processing unmodified audio collected during a time interval via at least one microphone with a first noise reduction model to obtain a first version of audio;
generating a first sequence of monitored media signatures based on the first version of the audio collected via the at least one microphone during the time interval;
processing the unmodified audio with a second noise reduction model to generate a modified version of the audio, the second noise reduction model different from the first noise reduction model;
generating, via execution of instructions with at least one processor, the modified version of the audio collected during the time interval, wherein the modified version differs from the first version;
generating a second sequence of monitored media signatures based on the modified version; and
reporting the first sequence of monitored media signatures and the second sequence of monitored media signatures.
18. The method of claim 17 , wherein the first version of the audio is the unmodified audio collected via the at least one microphone during the time interval, and wherein generating the modified version includes applying the second noise reduction model to the unmodified audio.
19. The method of claim 17 , further including:
processing the unmodified audio collected during the time interval via the at least one microphone with a first combination of noise reduction models to obtain the first version of the audio; and
processing the unmodified audio with a second combination of noise reduction models to generate the modified version of the audio, the second combination different from the first combination.