Conversation processing apparatus and conversation processing method

• US 10,013,982 B2
• Filed: 04/06/2016
• Issued: 07/03/2018
• Est. Priority Date: 04/16/2015
• Status: Active Grant

• Current Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Mizumoto, Takeshi, Nakadai, Kazuhiro, Utsunomiya, Kenta
• Primary Examiner(s): Sanchez, Michael Ortiz
• Application Number:
  US15/091,886
• Publication Number:
  US 20160307570A1
• Time in Patent Office: 818 Days
• CPC Class Codes: G10L 17/02 Preprocessing operations, e.g. segment selection; Pattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal components; Feature selection or extraction
• CPC Class Codes: G10L 17/06 Decision making techniques; Pattern matching strategies
• CPC Class Codes: G10L 25/21 the extracted parameters being power information
• CPC Class Codes: G10L 25/51 for comparison or discrimination

Abstract

A conversation processing apparatus includes a sound collection unit configured to collect voice signals of a plurality of speakers, a conversation recording unit configured to record the voice signals of the speakers collected by the sound collection unit, and a conversation analysis unit configured to calculate a degree to which a conversation between any two speakers is stimulated by analyzing the conversation between the two speakers from the voice signals recorded on the conversation recording unit.

Claims

1. A conversation processing apparatus comprising:

a sound collection unit configured to collect voice signals of a plurality of speakers;

a conversation recording unit configured to record the voice signals of the speakers collected by the sound collection unit;

a conversation analysis unit configured to calculate a degree to which a conversation between any two speakers is stimulated by analyzing the conversation between the two speakers from the voice signals recorded on the conversation recording unit; and

an analysis result output unit configured to output an analysis result based on the degree of stimulation calculated by the conversation analysis unit for display,

wherein the conversation analysis unit normalizes a volume based on a length of the voice signal spoken once for the voice signal in the conversation between any two speakers and decreases the degree of stimulation according to passage of time, and

wherein the conversation analysis unit decreases the degree of stimulation according to a formula shown below;

2. The conversation processing apparatus according to claim 1, wherein the degree of stimulation is based on an influence of each of any two speakers on the conversation in time series.

3. The conversation processing apparatus according to claim 1, wherein the conversation analysis unit calculates the degree of stimulation between any two speakers using a thermal equation.

4. The conversation processing apparatus according to claim 1, wherein the conversation analysis unit selects all pairs of any two speakers for the plurality of speakers, calculates the degree of stimulation for each selected pair, generates a weighted complete graph constituted of nodes and edges using the calculated degree of stimulation for each pair for weights of the edges, and estimates a central person of the conversation on the basis of the generated weighted complete graph,

wherein the central person of the conversation is the analysis result.

5. The conversation processing apparatus according to claim 4, further comprising a display unit which receives and displays the analysis result output from the analysis result output unit.

6. The conversation processing apparatus according to claim 1, wherein the conversation analysis unit calculates the degree of stimulation between any two speakers for every predetermined time.

7. The conversation processing apparatus according to claim 1, wherein the conversation analysis unit determines that the conversation is active when the degree of stimulation between any two speakers is greater than a predetermined threshold value.

8. The conversation processing apparatus according to claim 1, comprising:

a sound source localization unit configured to localize a sound source position using the voice signal collected by the sound collection unit; and

a sound source separation unit configured to perform sound source separation on the basis of a result of the localization by the sound source localization unit,

wherein the conversation recording unit records the voice signal of each speaker separated by the sound source separation unit.

10. The conversation processing apparatus according to claim 1, further comprising a display unit which receives and displays the analysis result output from the analysis result output unit.

9. A conversation processing method comprising:

collecting voice signals of a plurality of speakers;

recording the voice signals of the speakers collected in the sound collection procedure;

calculating a degree to which a conversation between any two speakers is stimulated by analyzing the conversation between the two speakers from the voice signals recorded in the conversation recording procedure;

normalizing a volume based on a length of the voice signal spoken once for the voice signal in the conversation between any two speakers and decreasing the degree of stimulation according to passage of time, wherein the degree of stimulation is decreased according to a formula shown below;

11. The conversation processing method according to claim 9, further comprising:

selecting all pairs of any two speakers for the plurality of speakers;

calculating the degree of stimulation for each selected pair;

generating a weighted complete graph constituted of nodes and edges using the calculated degree of stimulation for each pair for weights of the edges;

estimating a central person of the conversation on the basis of the generated weighted complete graph; and

displaying the central person of the conversation as the output.