SIGNAL PROCESSING APPARATUS AND SIGNAL PROCESSING METHOD

WIPO Patent Application WO/2011/105003

A1

Provided are a signal processing apparatus and a signal processing method whereby even in an ordinary environment, it can be precisely detected that a conversation has been completed. In a signal processing apparatus (100), a sound source separating unit (130) separates mixed sound signals, in which a plurality of sound sources are mixed, for respective sound sources. A speech detecting unit (140) performs a speech detection process for each of a plurality of sound source signals obtained by the separation; determines whether the respective sound source signal is a speech; and generates speech section information indicating speech/non-speech information for the respective sound source signal. A discrimination parameter extracting unit (150) extracts, based on the plurality of sound source signals or the speech section information, discrimination parameters indicating characteristic amounts of ordinary conversation. A conversation completion level calculating unit (160) calculates, based on the extracted discrimination parameters, and outputs a conversation completion level. A conversation partner determining unit (170) uses the conversation completion level to determine which sound source is a conversation partner.

YAMADA, Maki (())
山田麻紀 (())
ENDO, Mitsuru (())

JP2011/000358

September 01, 2011

January 24, 2011

Click for automatic bibliography generation  

PANASONIC CORPORATION (1006, Oaza Kadoma Kadoma-sh, Osaka 01, 〒5718501, JP)
パナソニック株式会社 (〒01 大阪府門真市大字門真１００６番地 Osaka, 〒5718501, JP)
YAMADA, Maki (())
山田麻紀 (())

G10L11/00; G10L15/04; G10L15/10; G10L21/02

WASHIDA, Kimihito (8th Floor, Shinjuku First West Bldg.1-23-7, Nishi-Shinjuku, Shinjuku-k, Tokyo 23, 〒1600023, JP)

View/Download PDF      PDF Help

Previous Patent: LITHIUM-ION SECONDARY BATTERY

Next Patent: PUPIL DETECTION DEVICE AND PUPIL DETECTION METHOD