Speech recognition apparatus

• US 8,060,368 B2
• Filed: 08/18/2006
• Issued: 11/15/2011
• Est. Priority Date: 12/07/2005
• Status: Active Grant

• Current Assignee: MEL NavIP LLC (Ni, Wang & Massand, PLLC)
• Original Assignee: Mitsubishi Electric Corporation
• Inventors: Ebihara, Takashi, Osawa, Masanobu, Okada, Reiko
• Primary Examiner(s): Azad; Abul K
• Application Number:
  US11/992,938
• Publication Number:
  US 20090228276A1
• Time in Patent Office: 1,915 Days
• Field of Search: 704/235, 704251-257, 704/270.1, 704/243, 704/10
• US Class Current: 704/257
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/228 of application context

Abstract

A voice recognition apparatus 10, which performs voice recognition of an input voice by referring to a voice recognition dictionary and outputs a voice recognition result, has an external information acquiring section 14 for acquiring from externally connected devices 20-1-20-N connected thereto a type of each externally connected device, and for acquiring data recorded in each externally connected device; a vocabulary extracting analyzing section 15 and 16 for extracting a vocabulary item from the data as an extracted vocabulary item, and for producing analysis data by analyzing the extracted vocabulary item and by providing the extracted vocabulary item with reading; and a dictionary generating section 17 for storing the analysis data in the voice recognition dictionary corresponding to the type. For each type of the externally connected devices, one of the voice recognition dictionaries 13-1-13-N is assigned.

Claims

1. A speech recognition apparatus comprising:

an external information acquiring section configured to acquire, from an externally connected device connected to the speech recognition apparatus, a device type of the externally connected device, and for acquiring data recorded in said externally connected device;

a vocabulary extracting analyzing section configured to extract a vocabulary item as an extracted vocabulary item form the data acquired by the external information acquiring section, and to produce analysis data by providing the extracted vocabulary item with pronunciation obtained through analysis of said extracted vocabulary item;

a dictionary generating section for configured to generate speech recognition dictionaries according to device types of externally connected devices by storing the analysis data produced by the vocabulary extracting analyzing section into a speech recognition dictionary corresponding to the device type acquired by the external information acquiring section; and

a speech recognition section configured to carry out speech recognition of input speech by referring to a speech recognition dictionary out of the speech recognition dictionaries generated by the dictionary generating section, and to output a result of said speech recognition, where the speech recognition dictionary to be used for said speech recognition corresponds to the device type acquired by the external information acquiring section.

2. The speech recognition apparatus according to claim 1, wherein said dictionary generating section comprises:

a type deciding section for configured to decide whether the speech recognition dictionary corresponding to the device type acquired by the external information acquiring section is present or not;

a dictionary creating section configured to newly create a speech recognition dictionary corresponding to the device type when a decision made by the type deciding section indicates negative, and to store the analysis data into the newly created speech recognition dictionary; and

a dictionary updating section configured to update the speech recognition dictionary corresponding to the device type by storing data of the analysis data into said speech recognition dictionary when a decision made by the type deciding section indicates positive, where the data to be stored is not present in said speech recognition dictionary.

3. The speech recognition apparatus according to claim 1, further comprising a speech recognition dictionary control section configured to activate only the speech recognition dictionary corresponding to the device type acquired by the external information acquiring section.

4. The speech recognition apparatus according to claim 1, wherein said vocabulary extracting analyzing section comprises:

an analysis dictionary used for analyzing the extracted vocabulary item; and

a resembling vocabulary deciding section configured to decide, when a vocabulary item matching the extracted vocabulary item is present in said analysis dictionary, as to whether a vocabulary item similar to the extracted vocabulary item is present in said analysis dictionary or not, wherein said speech recognition apparatus further comprises;

a vocabulary candidate presenting section for configured to exhibit the similar vocabulary item as a vocabulary candidate when a decision made by the resembling vocabulary deciding section indicates positive.

5. The speech recognition apparatus according to claim 4, further comprising a vocabulary candidate selecting section configured to recognize one of vocabulary candidates selected by a user of the speech recognition apparatus as a selected vocabulary item,

wherein said vocabulary extracting analyzing section produces analysis data by providing the selected vocabulary item with pronunciation of said selected vocabulary item.