Apparatus and method for clustering speakers, and a non-transitory computer readable medium thereof

Patent metadata

Field	Value
Publication number	US-9251808-B2
Application number	US-201213412694-A
Country	US
Kind code	B2
Filing date	Mar 6, 2012
Priority date	Jul 28, 2011
Publication date	Feb 2, 2016
Grant date	Feb 2, 2016

Abstract

According to one embodiment, a speaker clustering apparatus includes a clustering unit, an extraction unit, and an error detection unit. The clustering unit is configured to extract acoustic features for speakers from an acoustic signal, and to cluster utterances included in the acoustic signal into the speakers by using the acoustic features. The extraction unit is configured to acquire character strings representing contents of the utterances, and to extract linguistic features of the speakers by using the character strings. The error detection unit is configured to decide that, when one of the character strings does not fit with a linguistic feature of a speaker into which an utterance of the one is clustered, the utterance is erroneously clustered by the clustering unit.

First claim

What is claimed is:

1. An apparatus for clustering speakers, the apparatus comprising:
a memory device that stores an acoustic signal already recorded:
a clustering unit executed by a computer using a stored program, configured to extract acoustic features for speakers from the acoustic signal, and to cluster utterances included in the acoustic signal into each of the speakers by using the acoustic features;
an acquisition unit executed by the computer, configured to acquire character strings by recognizing the acoustic signal of each utterance;
an extraction unit executed by the computer, configured to acquire the character strings representing contents of the utterances, and to extract a linguistic feature of each of the speakers by comparing the character strings with a first person rule and an end of sentence rule;
an error detection unit executed by the computer, configured to decide that, when one of the character strings does not fit with the linguistic feature of a speaker into which an utterance of the one is clustered, the utterance is erroneously clustered by the clustering unit; and
a display that displays at least one of a clustering result of the clustering unit and a decision result of the error detection unit.

2. The apparatus according to claim 1 , further comprising:
a re-clustering unit executed by the computer, configured to decide that, when the one fits with the linguistic feature of another speaker, the utterance is to be clustered into the another speaker.

3. The apparatus according to claim 1 , wherein
the extraction unit decides whether the first person rule and the end of sentence rule previously-stored fit with the character strings of the utterances clustered into each of the speakers, and sets the rule of which the number of utterances of fitted character strings is larger than a predetermined threshold, as the linguistic feature of the speaker into which the utterances of fitted character strings are clustered.

4. The apparatus according to claim 2 , wherein
the display displays a decision result of the re-clustering unit.

5. A method for clustering speakers, the method being realized by executing a stored program by a computer, the comprising:
storing by the computer into a memory device, acoustic signal already recorded;
extracting, by the computer, acoustic features for speakers from the acoustic signal;
clustering, by the computer, utterances included in the acoustic signal into each of the speakers by using the acoustic features;
acquiring, by the computer, character strings by recognizing the acoustic signal of each utterance;
acquiring, by the computer, the character strings representing contents of the utterances;
extracting, by the computer, a linguistic feature of each of the speakers by comparing the character strings with a first person rule and an end of sentence rule;
deciding, by the computer, that, when one of the character strings does not fit with the linguistic feature of a speaker into which an utterance of the one is clustered, the utterance is erroneously clustered by the clustering; and
displaying by the computer via a display, at least one of a clustering result of the clustering and a decision result of the deciding.

6. A non-transitory computer readable medium for causing a computer to perform a method for clustering speakers, the method comprising:
storing, by the computer into a memory device, acoustic signal already recorded;
extracting acoustic features for speakers from an acoustic signal;
clustering utterances included in the acoustic signal into each of the speakers by using the acoustic features;
acquiring character strings by recognizing the acoustic signal of each utterance;
acquiring the character strings representing contents of the utterances;
extracting a linguistic feature of each of the speakers by comparing the character strings with a first person rule and an end of sentence rule;
deciding that, when one of the character strings does not fit with the linguistic feature of a speaker into which an utterance of the one is clustered, the utterance is erroneously clustered by the clustering; and
displaying by the computer via a display, at least one of a clustering result of the clustering and a decision result of the deciding.

Assignees

• Ikeda Tomoo
• Nagao Manabu
• Nishiyama Osamu
• Suzuki Hirokazu
• Ueno Koji
• Shimogori Nobuhiro
• Toshiba Kk

Inventors

• Ikeda Tomoo
• Nagao Manabu
• Nishiyama Osamu
• Suzuki Hirokazu
• Ueno Koji
• Shimogori Nobuhiro

Classifications

• G10L21/028 Primary

  using properties of sound source · CPC title

• G10L15/26

  Speech to text systems (G10L15/08 takes precedence) · CPC title