Methods and apparatus to determine audio source impact on an audience of media

Patent metadata

Field	Value
Publication number	US-10595083-B2
Application number	US-201816140238-A
Country	US
Kind code	B2
Filing date	Sep 24, 2018
Priority date	Apr 20, 2018
Publication date	Mar 17, 2020
Grant date	Mar 17, 2020

Abstract

Methods, apparatus, systems and articles of manufacture to determine audio source impact on an audience of media are disclosed. A disclosed example method includes dividing monitored audio into successive audio segments including a first audio segment and a second audio segment, the monitored audio associated with monitored media. The example method also includes generating a first confidence value for the first audio segment of the successive segments and a second confidence value for the second audio segment, the first confidence value associated with a presence of a first audio source in the first audio segment, the second confidence value associated with a presence of the first audio source in the second audio segment. The example method further includes comparing (1) a first correlation of the first confidence value and a first rating associated with the first audio segment and (2) a second correlation of the second confidence value and a second rating associated with the second audio segment to determine an impact of the first audio source on audience ratings of the monitored media.

First claim

What is claimed is:

1. An apparatus comprising:
an audio segmenter implemented by hardware or at least one processor to divide monitored audio into successive audio segments including a first audio segment and a second audio segment, the monitored audio associated with monitored media;
a speaker evaluator to generate a first confidence value from the first audio segment of the successive segments and a second confidence value from the second audio segment, the first confidence value associated with a presence of a voice of a first person in the first audio segment, the second confidence value associated with a presence of the voice of the first person in the second audio segment;
a sentiment analyzer to:
identify a first dialogue associated with the first person and the first audio segment;
identify a second dialogue associated with the first person and the second audio segment; and
determine a first sentiment associated with the first dialogue and a second sentiment associated with the second dialogue; and

an audience rating correlator to:
augment a first rating with the first sentiment, the first rating associated with the first audio segment;
augment a second rating with the second sentiment, the second rating associated with the second audio segment; and
compare (1) a first correlation of the first confidence value and the first rating associated with the first audio segment, and (2) a second correlation of the second confidence value and the second rating associated with the second audio segment to determine an impact of the voice of the first person on audience ratings of the monitored media.

2. The apparatus of claim 1 , wherein the speaker evaluator includes:
a speech pattern generator to generate a pattern from the first audio segment; and
a pattern comparator to compare the generated pattern to a set of reference patterns to generate the first confidence value, the set of reference patterns associated with the voice of the first person.

3. The apparatus of claim 2 , wherein the pattern comparator is to compare the generated pattern to the set of reference patterns based on linear regression.

4. The apparatus of claim 1 , wherein the voice of the first person is a first audio source and the speaker evaluator is further to generate a third confidence value from the first audio segment, the third confidence value associated with a presence of a second audio source in the first audio segment, and further includes a dominant speaker identifier to compare the first confidence value and the third confidence value to determine a dominant audio source in the first segment.

5. The apparatus of claim 4 , wherein the second audio source corresponds to a reference sound.

6. A method comprising:
dividing, by executing an instruction with a processor, monitored audio into successive audio segments including a first audio segment and a second audio segment, the monitored audio associated with monitored media;
generating, by executing an instruction with the processor, a first confidence value from the first audio segment and a second confidence value from the second audio segment, the first confidence value associated with a presence of a voice of a first person in the first audio segment, the second confidence value associated with a presence of the voice of the first person in the second audio segment;
identifying, by executing an instruction with the processor, a first dialogue associated with the first person and the first audio segment;
identifying, by executing an instruction with the processor, a second dialogue associated with the first person and the second audio segment;
determining, by executing an instruction with the processor, a first sentiment associated with the first dialogue and a second sentiment associated with the second dialogue;
augmenting, by executing an instruction with the processor, a first rating with the first sentiment, the first rating associated with the first audio segment;
augmenting, by executing an instruction with the processor, a second rating with the first sentiment, the second rating associated with the second audio segment; and
comparing, by executing an instruction with the processor, (1) a first correlation of the first confidence value and the first rating associated with the first audio segment, and (2) a second correlation of the second confidence value and the second rating associated with the second audio segment to determine an impact of the voice of the first person on audience ratings of the monitored media.

7. The method of claim 6 , wherein the generating of the first confidence value includes:
generating a pattern from the first audio segment; and
comparing the generated pattern to a set of reference patterns to generate the first confidence value, the set of reference patterns associated with the voice of the first person.

8. The method of claim 7 , wherein the comparison of the generated pattern to the set of reference patterns is based on linear regression.

9. The method of claim 6 , wherein the voice of the first person is a first audio source, and further including:
generating a third confidence value from the first audio segment, the third confidence value associated with a presence of a second audio source in the first audio segment; and
comparing the first confidence value and the third confidence value to determine a dominant audio source in the first segment.

10. The method of claim 6 , wherein the identifying of the first dialogue includes analyzing closed captioning data associated with the monitored media.

11. A non-transitory computer readable medium comprising instructions that, when executed, cause a processor to at least:
divide monitored audio into successive audio segments including a first audio segment and a second audio segment, the monitored audio associated with monitored media;
generate a first confidence value from the first audio segment and a second confidence value from the second audio segment, the first confidence value associated with a presence of a voice of a first person in the first audio segment, the second confidence value associated with a presence of the voice of the first person in the second audio segment;
identify a first dialogue associated with the first person and the first audio segment;
identify a second dialogue associated with the first person and the second audio segment;
determine a first sentiment associated with the first dialogue and a second sentiment associated with the second dialogue;
augment a first rating with the first sentiment, the first rating associated with the first audio segment;
augment a second rating with the first sentiment, the second rating associated with the second audio segment; and
compare (1) a first correlation of the first confidence value and the first rating associated with the first audio segment, and (2) a second correlation of the second confidence value and the second rating associated with the second audio segment to determine an impact of the first voice of the first person on audience ratings of the monitored media.

12. The non-transitory computer readable medium of claim 11 , wherein the instructions, when executed, cause the processer to:
generate a pattern from the first audio segment; and
compare the generated pattern to a set of reference patterns to generate the first confidence value, the set of reference patterns associated with the voice of the first person.

13. The non-transitory computer readable medium of claim 12 , wherein instructions, when executed, cause the processor to compare the generated pattern to the set of reference patterns based on linear regression.

Assignees

• Nielsen Co Us Llc

Inventors

• Milavsky Joseph Gregory
• Alsaket Hashem
• Bartelme James
• Besehanic Jan
• Stembler Edward

Classifications

• H04H60/58

  of audio {(determination or detection of speech characteristics in general G10L25/00; speech recognition in general G10L15/00)} · CPC title

• G10L17/00

  Speaker identification or verification techniques · CPC title

• H04H60/33

  Arrangements for monitoring the users' behaviour or opinions · CPC title

• H04N21/252

  Processing of multiple end-users' preferences to derive collaborative data · CPC title

• H04N21/42203

  sound input device, e.g. microphone · CPC title