Back to results
US 2002/0120447 A1Published application
Speech processing system
AssigneeCanon Kabushiki Kaisha
InventorsJason Peter Andrew Charlesworth, Jebu Jacob Rajan
Life of this patent
Estimating the term…
Abstract
A system is provided for allowing a user to add word models to a speech recognition system. In particular, the system allows a user to input a number of renditions of the new word and which generates from these a sequence of phonemes representative of the new word. This representative sequence of phonemes is stored in a word to phoneme dictionary together with the typed version of the word for subsequent use by the speech recognition system.
Claims62
An apparatus for generating a sequence of sub-word units representative of a new word to be added to a dictionary of a speech recognition system, the apparatus comprising:
- first receiving means for receiving signals representative of first and second spoken renditions of the new word;
- speech recognition means for comparing the received first and second spoken renditions with pre-stored sub-word unit models to generate first and second sequences of sub-word units representative of said first and second spoken renditions of the new word respectively;
- means for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- and
- means for determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning means.
2 more elements
An apparatus for generating a sequence of sub-word units representative of a new word to be added to a dictionary of a speech recognition system, the apparatus comprising:
- means for receiving signals representative of a plurality of spoken renditions of the new word;
- speech recognition means for comparing the received spoken renditions with pre-stored sub-word unit models to generate a corresponding plurality of sequences of sub-word units representative of said plurality of spoken renditions;
- means for aligning sub-word units of each rendition with sub-word units of units of other renditions to form a number of aligned groups of sub-word units, each aligned group comprising a sub-word unit from each rendition;
- and
- means for determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned groups of sub-word units determined by said aligning means.
2 more elements
An apparatus for adding a new word and sub-word representation of the new word to a word dictionary of a speech recognition system, the apparatus comprising,
- means for receiving a first sequence of sub-word units representative of a first spoken rendition of the new word and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new word;
- means for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- means for determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning means;
- and
- means for adding the new word and the representative sequence of sub-word units to said word dictionary.
2 more elements
A speech recognition system comprising:
- means for receiving speech signals to be recognised;
- means for storing sub-word unit models;
- means for comparing received speech with the sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals;
- a word dictionary relating sequences of sub-word units to words;
- a word decoder for processing the one or more sequences of sub-word units output by said comparing means using the word dictionary to generate one or more words corresponding to the received speech signals;
- an apparatus for adding a new word and a sub-word representation of the new word to the word dictionary;
- and
- means for controllably connecting the output of said comparing means to either said word decoder or said apparatus for adding the new word and a sub-word representation of the new word to the word dictionary;
- characterised in that said apparatus for adding the new word and a sub-word representation of the new word to the word dictionary comprises:
- means for receiving a first sequence of sub-word units representative of a first spoken rendition of the new word output by said comparing means and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new word output by said comparing means;
- means for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- means for determining sequences of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning means;
- means for receiving a text rendition of the new word;
- and
- means for adding said text rendition of the new word and the representative sequence of sub-word units to said word dictionary.
12 more elements
A speech recognition system comprising:
- means for receiving speech signals to be recognised;
- means for storing sub-word unit models;
- means for comparing received speech with the sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals;
- a command dictionary relating sequences of sub-word units to commands;
- and
- a command decoder for processing the one or more sequences of sub-word units output by said comparing means using the command dictionary to generate one or more commands corresponding to the received speech signals;
- an apparatus for adding a new command and a sub-word representation of the new command to the command dictionary;
- and
- means for controllably connecting the output of said comparing means to either said command decoder or said apparatus for adding the new command and a sub-word representation of the new command to the command dictionary;
- characterised in that said apparatus for adding the new command and a sub-word representation of the new command to the command dictionary comprises:
- means for receiving a first sequence of sub-word units representative of a first spoken rendition of the new command output by said comparing means and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new command output by said comparing means;
- means for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- means for determining sequences of sub-word units representative of the spoken renditions of the new command in dependence upon the aligned pairs of sub-word units determined by said aligning means;
- and
- means for adding said representative sequence of sub-word units to said command dictionary together with the corresponding new command.
12 more elements
A method of generating a sequence of sub-word units representative of a new word to be added to a dictionary of a speech recognition system, the method comprising the steps of:
- a first receiving step of receiving signals representative of first and second spoken renditions of the new word;
- comparing the received first and second spoken renditions with pre-stored sub-word unit models to generate a first sequence of sub-word units representative of said first spoken rendition of the new word and a second sequence of sub-word units representative of said second spoken rendition of the new word;
- aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- and
- determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step.
2 more elements
A method of generating a sequence of sub-word units representative of a new word to be added to a dictionary of a speech recognition system, the method comprising the steps of:
- receiving signals representative of a plurality of spoken renditions of the new word;
- comparing the received spoken renditions with pre-stored sub-word unit models to generate a corresponding plurality of sequences of sub-word units representative of said plurality of spoken renditions;
- aligning sub-word units of each rendition with sub-word units of units of other renditions to form a number of aligned groups of sub-word units, each aligned group comprising a sub-word unit from each rendition;
- and
- determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned groups of sub-word units determined by said aligning step.
2 more elements
A method of adding a new word and sub-word representation of the new word to a word dictionary of a speech recognition system, the method comprising the steps of:
- receiving a first sequence of sub-word units representative of a first spoken rendition of the new word and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new word;
- aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step;
- and
- adding the new word and the representative sequence of sub-word units to said word dictionary.
2 more elements
A speech recognition method comprising the steps of:
- receiving speech signals to be recognised;
- storing sub-word unit models;
- comparing received speech signals with the sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals;
- storing a word dictionary relating sequences of sub-word units to words;
- processing the one or more sequences of sub-word units output by said comparing step using the stored word dictionary to generate one or more words corresponding to the received speech signals;
- the step of adding a new word and a sub-word representation of the new word to the word dictionary;
- and
- controllably feeding the output of said comparing step to either said processing step or said adding step;
- characterised in that said adding step comprises:
- receiving a first sequence of sub-word units representative of a first spoken rendition of the new word output by said comparing step and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new word output by said comparing step;
- aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- determining sequences of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step;
- receiving a text rendition of the new word;
- and
- adding said text rendition of the new word and the representative sequence of sub-word units to said word dictionary.
12 more elements
A speech recognition method comprising the steps of:
- receiving speech signals to be recognised;
- storing sub-word unit models;
- comparing received speech signals with the sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals;
- storing a command dictionary relating sequences of sub-word units to commands;
- processing the one or more sequences of sub-word units output by said comparing step using the stored command dictionary to generate one or more commands corresponding to the received speech signals;
- adding a new command and a sub-word representation of the new command to the command dictionary;
- and
- controllably feeding the output of said comparing step to either said processing step or said adding step;
- characterised in that said adding step comprises:
- receiving a first sequence of sub-word units representative of a first spoken rendition of the new command output by said comparing step and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new command output by said comparing step;
- aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- determining sequences of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step;
- and
- adding said sequence of sub-word units to said command dictionary together with the associated new command.
11 more elements
A storage medium storing processor implementable instructions for controlling a processor to carry out a method of generating a sequence of sub-word units representative of a new word to be added to a dictionary of a speech recognition system, the processor instructions comprising:
- first receiving instructions for receiving signals representative of first and second spoken renditions of the new word;
- instructions for comparing the received first and second spoken renditions with pre-stored sub-word unit models to generate a first sequence of sub-word units representative of said first spoken rendition of the new word and a second sequence of sub-word units representative of said second spoken rendition of the new word;
- instructions for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- and
- instructions for determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step.
2 more elements
A storage medium storing processor implementable instructions for controlling a processor to carry out a method of generating a sequence of sub-word units representative of a new word to be added to a dictionary of a speech recognition system, the processor instructions comprising:
- instructions for receiving signals representative of a plurality of spoken renditions of the new word;
- instructions for comparing the received spoken renditions with pre-stored sub-word unit models to generate a corresponding plurality of sequences of sub-word units representative of said plurality of spoken renditions;
- instructions for aligning sub-word units of each rendition with sub-word units of units of other renditions to form a number of aligned groups of sub-word units, each aligned group comprising a sub-word unit from each rendition;
- and
- instructions for determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned groups of sub-word units determined by said aligning step.
2 more elements
A storage medium storing processor implementable instructions for controlling a processor to carry out a method of adding a new word and sub-word representation of the new word to a word dictionary of a speech recognition system, the process instructions comprising:
- instructions for receiving a first sequence of sub-word units representative of a first spoken rendition of the new word and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new word;
- instructions for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- instructions for determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step;
- and
- instructions for adding the new word and the representative sequence of sub-word units to said word dictionary.
2 more elements
A storage medium storing processor implementable instructions for controlling a processor to carry out a speech recognition method, the process instructions comprising:
- instructions for receiving speech signals to be recognised;
- instructions for storing sub-word unit models;
- instructions for comparing received speech signals with the sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals;
- instructions for storing a word dictionary relating sequences of sub-word units to words;
- instructions for processing the one or more sequences of sub-word units output by said comparing step using the stored word dictionary to generate one or more words corresponding to the received speech signals;
- instructions for adding a new word and a sub-word representation of the new word to the word dictionary;
- and
- instructions for controllably feeding the output of said comparing step to either said processing step or said adding step;
- characterised in that said adding instructions comprise:
- instructions for receiving a first sequence of sub-word units representative of a first spoken rendition of the new word output by said comparing step and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new word output by said comparing step;
- instructions for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- instructions for determining sequences of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step;
- instructions for receiving a text rendition of the new word;
- and
- instructions for adding said text rendition of the new word and the representative sequence of sub-word units to said word dictionary.
12 more elements
A storage medium storing processor implementable instructions for controlling a processor to carry out a speech recognition method, the process instructions comprising:
- instructions for receiving speech signals to be recognised;
- instructions for storing sub-word unit models;
- instructions for comparing received speech signals with the sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals;
- instructions for storing a command dictionary relating sequences of sub-word units to commands;
- instructions for processing the one or more sequences of sub-word units output by said comparing step using the stored command dictionary to generate one or more commands corresponding to the received speech signals;
- instructions for adding a new command and a sub-word representation of the new command to the command dictionary;
- and
- instructions for controllably feeding the output of said comparing step to either said processing step or said adding step;
- characterised in that said adding instructions comprise:
- instructions for receiving a first sequence of sub-word units representative of a first spoken rendition of the new command output by said comparing step and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new command output by said comparing step;
- instructions for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- instructions for determining sequences of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step;
- and
- instructions for adding said sequence of sub-word units to said command dictionary together with the associated new command.
11 more elements
Processor implementable instructions for controlling a processor to carry out a method of generating a sequence of sub-word units representative of a new word to be added to a dictionary of a speech recognition system, the processor instructions comprising:
- first receiving instructions for receiving signals representative of first and second spoken renditions of the new word;
- instructions for comparing the received first and second spoken renditions with pre-stored sub-word unit models to generate a first sequence of sub-word units representative of said first spoken rendition of the new word and a second sequence of sub-word units representative of said second spoken rendition of the new word;
- instructions for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- and
- instructions for determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step.
2 more elements
Processor implementable instructions for controlling a processor to carry out a method of generating a sequence of sub-word units representative of a new word to be added to a dictionary of a speech recognition system, the processor instructions comprising:
- instructions for receiving signals representative of a plurality of spoken renditions of the new word;
- instructions for comparing the received spoken renditions with pre-stored sub-word unit models to generate a corresponding plurality of sequences of sub-word units representative of said plurality of spoken renditions;
- instructions for aligning sub-word units of each rendition with sub-word units of units of other renditions to form a number of aligned groups of sub-word units, each aligned group comprising a sub-word unit from each rendition;
- and
- instructions for determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned groups of sub-word units determined by said aligning step.
2 more elements
Processor implementable instructions for controlling a processor to carry out a method of adding a new word and sub-word representation of the new word to a word dictionary of a speech recognition system, the process instructions comprising:
- instructions for receiving a first sequence of sub-word units representative of a first spoken rendition of the new word and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new word;
- instructions for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- instructions for determining a sequence of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step;
- and
- instructions for adding the new word and the representative sequence of sub-word units to said word dictionary.
2 more elements
A storage medium storing processor implementable instructions for controlling a processor to carry out a speech recognition method, the process instructions comprising:
- instructions for receiving speech signals to be recognised;
- instructions for storing sub-word unit models;
- instructions for comparing received speech signals with the sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals;
- instructions for storing a word dictionary relating sequences of sub-word units to words;
- instructions for processing the one or more sequences of sub-word units output by said comparing step using the stored word dictionary to generate one or more words corresponding to the received speech signals;
- instructions for adding a new word and a sub-word representation of the new word to the word dictionary;
- and
- instructions for controllably feeding the output of said comparing step to either said processing step or said adding step;
- characterised in that said adding instructions comprise:
- instructions for receiving a first sequence of sub-word units representative of a first spoken rendition of the new word output by said comparing step and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new word output by said comparing step;
- instructions for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- instructions for determining sequences of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step;
- instructions for receiving a text rendition of the new word;
- and
- instructions for adding said text rendition of the new word and the representative sequence of sub-word units to said word dictionary.
12 more elements
A storage medium storing processor implementable instructions for controlling a processor to carry out a speech recognition method, the process instructions comprising:
- instructions for receiving speech signals to be recognised;
- instructions for storing sub-word unit models;
- instructions for comparing received speech signals with the sub-word unit models to generate one or more sequences of sub-word units representative of the received speech signals;
- instructions for storing a command dictionary relating sequences of sub-word units to commands;
- instructions for processing the one or more sequences of sub-word units output by said comparing step using the stored command dictionary to generate one or more commands corresponding to the received speech signals;
- instructions for adding a new command and a sub-word representation of the new command to the command dictionary;
- and
- instructions for controllably feeding the output of said comparing step to either said processing step or said adding step;
- characterised in that said adding instructions comprise:
- instructions for receiving a first sequence of sub-word units representative of a first spoken rendition of the new command output by said comparing step and for receiving a second sequence of sub-word units representative of a second spoken rendition of the new command output by said comparing step;
- instructions for aligning sub-word units of the first sequence with sub-word units of the second sequence to form a number of aligned pairs of sub-word units;
- instructions for determining sequences of sub-word units representative of the spoken renditions of the new word in dependence upon the aligned pairs of sub-word units determined by said aligning step;
- and
- instructions for adding said sequence of sub-word units to said command dictionary together with the associated new command.
11 more elements
Dependencies are inferred from the claim text.