Method and system for laughter detection
First Claim
1. A method for detecting an at least one laughter episode in an audio signal of a telephone interaction in a call center, the method comprising the steps of:
- capturing an audio signal of a telephone interaction in a call center;
detecting, an at least one burst in the audio signal;
detecting, at least predetermined number of consecutive bursts as N-tuple;
extracting in time domain an at least one time domain feature from the at least predetermined number of consecutive bursts, wherein the at least one time domain feature is selected from the group consisting of;
mean duty cycle, mean distance between voiced sections, and an itakura distance;
scoring the least one time domain feature against an at least one predetermined threshold;
determining whether the at least one feature complies with the at least one predetermined threshold; and
if the at least one time domain feature complies with the at least one predetermined threshold, determining whether the at least predetermined number of consecutive bursts represent a laughter episode.
3 Assignments
0 Petitions
Accused Products
Abstract
A method and apparatus for laughter detection. Laughter is detected through the presence of a sequence of at least a predetermined number such as three consecutive bursts, each burst comprising a voiced portion and an unvoiced portion. After detecting bursts, n-tuples such as triplets are detected, and a likelihood of each burst N-tuple to represent laughter is provided by comparison to predetermined thresholds. Finally, a total score is assigned to the signal based on the grades associated with the triplets and parameters such as the distance between the N-tuples, the total score representing the probability that the audio signal comprises a laughter episode. The method and apparatus preferably comprise a training step and module for determining the thresholds according to manually marked audio signals.
76 Citations
30 Claims
-
1. A method for detecting an at least one laughter episode in an audio signal of a telephone interaction in a call center, the method comprising the steps of:
-
capturing an audio signal of a telephone interaction in a call center; detecting, an at least one burst in the audio signal; detecting, at least predetermined number of consecutive bursts as N-tuple; extracting in time domain an at least one time domain feature from the at least predetermined number of consecutive bursts, wherein the at least one time domain feature is selected from the group consisting of;
mean duty cycle, mean distance between voiced sections, and an itakura distance;scoring the least one time domain feature against an at least one predetermined threshold; determining whether the at least one feature complies with the at least one predetermined threshold; and if the at least one time domain feature complies with the at least one predetermined threshold, determining whether the at least predetermined number of consecutive bursts represent a laughter episode. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 22)
-
-
21. An apparatus for detecting an at least one laughter episode in an audio signal of a telephone interaction captured in a call center, the apparatus comprising:
-
a burst detection component for detecting, an at least one burst in the audio signal; a burst N-tuple detection component for detecting, at least predetermined number of consecutive bursts; a burst N-tuple feature extraction component for extracting in time domain an at least one time domain feature from the at least predetermined number of consecutive bursts, wherein the at least one time domain feature is selected from the group consisting of;
mean duty cycle, mean distance between voiced sections, and an itakura distance;a scoring component for scoring the least one time domain feature against an at least one predetermined threshold; and a decision component for determining whether the at least one time domain feature is associated with a laughter episode, according to whether the at least one time domain feature complies with the at least one predetermined threshold. - View Dependent Claims (23, 24, 25, 26, 27, 28, 29, 30)
-
Specification