US 9,437,190 B2Granted

Speech recognition apparatus for recognizing user's utterance

AssigneeAsahi Kasei Kabushiki Kaisha

InventorsTomoyuki Kumai, Toshiyuki Miyazaki

Life of this patent

Estimating the term…

Loading patent…

Abstract

In accordance with alphabet input method information for each user, a word formed of an alphabet string is registered in a word dictionary, in a state where “dotto” being added before each alphabet and one of a set of alphabets difficult to distinguish from each other like “M and N” and “B and P” is repeated twice. For example, a word “PAM” and a feature of time series corresponding to “dotto P P doddo A dotto M” are registered in association with each other. When a user performs a speech input of “PAM”, in accordance with the user's alphabet input method information, the user utters “dotto P P dotto A dotto M”. A speech recognition is performed on this sound data using the word dictionary corresponding to the user's alphabet input method information.

Claims12

Dependent claims

A speech recognition apparatus operated by a computer for recognizing an utterance by a user, comprising:

a user information processor that receives user information inputted into the computer from an external device and generates user input method information representing an utterance method of the user;
a sound input device that is connected to the computer to receive a successive character string uttered by the user and converts the successive character string into sound data formed of digital signals;
a recognition dictionary;
a speech recognition processor that receives the utterance method of the user from the user information processor and the sound data from the sound input device to perform a speech recognition process using the recognition dictionary, wherein:
the speech recognition processor requests a recognition data that is registered in the recognition dictionary and corresponds to the utterance method of the user, the recognition data including a target character to be recognized and information on the sound data in association with each other, the sound data representing a pronunciation of the successive character string including a predetermined character string and the target character subsequent to the predetermined character string,
the speech recognition processor recognizes the target character from the sound data using a feature of acoustic model representing the target character, and
the predetermined character string is a Japanese character string which enhances sound energy of a first phoneme of the target character when the successive character string is uttered, and pronounces as having three or more moras including “tto” as last two moras;
and
an output device that outputs the target character recognized by the speech recognition processor to a display device of the computer.

6 more elements

A speech recognition apparatus operated by a computer for recognizing an utterance by a user, comprising:

a user information processor that receives user information inputted into the computer from an external device and generates user input method information representing an utterance method of the user;
a sound input device that is connected to the computer to receive a successive character string uttered by the user and convert the successive character string into sound data formed of digital signals;
a recognition dictionary;
and
a speech recognition processor that receives the utterance method of the user from the user information processor and the sound data from the sound input device so as to perform a speech recognition process using the recognition dictionary, wherein:
the speech recognition processor requests a recognition data that is registered in the recognition dictionary and corresponds to the utterance method of the user, the recognition data including a target character to be recognized and information on the sound data in association with each other, the sound data representing a pronunciation of the successive character string including a predetermined character string and the target character subsequent to the predetermined character string,
the speech recognition processor recognizes the target character from the sound data using a feature of acoustic model representing the target character, and
the predetermined character string is an English character string which enhances sound energy of a first phoneme of the target character when the successive character string is uttered, and pronounces as being formed of a consonant, a vowel, and a consonant in this order;
and
an output device that outputs the target character recognized by the speech recognition processor to a display device of the computer.

7 more elements

A speech recognition apparatus operated by a computer for recognizing an utterance by a user, comprising:

a user information processor that receives user information inputted into the computer from an external device and generates user input method information representing an utterance method of the user;
a sound input device that is connected to the computer to receive a successive character string uttered by the user and convert the successive character string into sound data formed of digital signals;
a recognition dictionary;
and
a speech recognition processor that receives the utterance method of the user from the user information processor and the sound data from the sound input device so as to perform a speech recognition process using the recognition dictionary, wherein:
the speech recognition processor requests a recognition data that is registered in the recognition dictionary and corresponds to the utterance method of the user, the recognition data including a target character to be recognized and information on the sound data in association with each other, the sound data representing a pronunciation of the successive character string including a predetermined character string and the target character subsequent to the predetermined character string,
the speech recognition processor recognizes the target character from the sound data using a feature of acoustic model representing the target character, and
the predetermined character string is a Chinese character string which enhances sound energy of a first phoneme of the target character when the successive character string is uttered, and pronounces as being classified into a third tone of Chinese tones and has a single syllable;
and
an output device that outputs the target character recognized by the speech recognition processor to a display device of the computer.

7 more elements

Dependencies are inferred from the claim text.

Facts

Application: 14/239,315Filed 31 Aug 2012
Priority: JP 2011-1976259 Sep 2011
Published: US 2014/0163987 A112 Jun 2014
Granted: 6 Sep 2016
Term ends: Not estimatedThe source records don’t support an estimate.

Classification3 codes

G10L 15/187
Using context dependencies, e.g. language models: phonemic context, e.g. pronunciation rules, phonotactical constraints or phoneme n-grams
G10L 2015/086
Speech classification or search: recognition of spelled words
G10L
Speech analysis techniques or speech synthesis

Family

US 2014/0163987 A1Pre-grant publication · published 12 Jun 2014
US 9,437,190 B2This patent

From the US filing history. Foreign family members aren’t listed.

Timeline

Citations

Open elsewhere

Building an agent or pulling many patents? Use the API or MCP server instead of scraping pages. Get an API key