Method and apparatus for a parameter sharing speech recognition system

An apparatus for speech recognition comprising:

• an input for receiving speech signals into a processor;
• a processor configured to recognize the received speech signals using a speech recognition system to generate a signal representative of the received speech signal, the speech recognition system produced by generating and training a plurality of phoneme models, wherein at least one of the plurality of phoneme models are shared among a plurality of phonemes, and at least a first one of the plurality of phoneme models are shared with at least a second one of the plurality of phoneme models;
• and
• an output for providing a signal representative of the received speech signal.

A speech recognition process comprising a statistical learning technique that uses a model, the model produced by:

• generating and training a plurality of phoneme models, wherein at least one of the plurality of phoneme models are shared among a plurality of phonemes;
• generating a plurality of shared probability sub-distribution functions from the trained plurality of phoneme models;
• and
• evaluating the plurality of phoneme models for further sharing in response to the plurality of shared probability sub-distribution functions.

A method for generating a plurality of phoneme models for use in a speech recognition system, the method comprising the steps of:

• retaining as a separate phoneme model a triphone phoneme model for which a number of trained frames exceeds a threshold;
• generating at least one shared phoneme model to represent a plurality of triphone phoneme models for which the number of trained frames having a common biphone exceeds the threshold;
• generating at least one shared phoneme model to represent a plurality of triphone phoneme models for which the number of trained frames having an equivalent effect on a phonemic context exceeds the threshold;
• and
• generating at least one shared phoneme model to represent a plurality of triphone phoneme models having the same center context.

A computer readable medium containing executable instructions which, when executed in a processing system, causes the system to perform the steps for recognizing speech comprising:

• receiving speech signals into a processor;
• processing the received speech signals using a speech recognition system comprising a plurality of phoneme models, wherein at least one of the plurality of phoneme models are shared among a plurality of phonemes and at least a first one of the plurality of phoneme models are shared with at least a second one of the plurality of phoneme models;
• and
• generating signals representative of the received speech signals.

A method for recognizing speech comprising the steps of:

• receiving speech signals into a processor;
• processing the received speech signals using a model comprising a plurality of phoneme models, wherein at least one of the plurality of phoneme models are shared among a plurality of phonemes, and at least a first one of the plurality of phoneme models are shared with at least a second one of the plurality of phoneme models;
• and
• generating signals representative of the received speech signals.

An apparatus for speech recognition comprising:

• an input configured to receive speech signals into a processor;
• a processor configured to process the received speech signals using a model comprising a plurality of phoneme models, wherein at least one of the plurality of phoneme models are shared among a plurality of phonemes, and at least a first one of the plurality of phoneme models are shared with at least a second one of the plurality of phoneme models;
• and
• an output configured to provide a signal representative of the received speech signal.

A computer readable medium containing executable instructions which, when executed in a processing system, causes the system to perform the steps for recognizing speech comprising:

• receiving speech signals into a processor;
• processing the received speech signals using a model comprising a plurality of context dependent phoneme models, wherein at least one of the plurality of phoneme models are shared among a plurality of phonemes, and at least a first one of the plurality of phoneme models are shared with at least a second one of the plurality of phoneme models;
• and
• providing output signals representative of the received speech signals.

A system for recognizing speech comprising:

• means for receiving speech signals into a processor;
• means for processing the received speech signals using a speech recognition system produced by generating a plurality of phoneme models, wherein at least one of the plurality of phoneme models are shared among a plurality of phonemes, and at least a first one of the plurality of phoneme models are shared with at least a second one of the plurality of phoneme models;
• and
• means for generating signals representative of the received speech signals.

A system for generating a plurality of phoneme models for use in a speech recognition system, comprising:

• means for retaining as a separate phoneme model a triphone phoneme model for which a number of trained frames exceeds a threshold;
• means for generating at least one shared phoneme model to represent a plurality of triphone phoneme models for which the number of trained frames having a common biphone exceeds the threshold;
• means for generating at least one shared phoneme model to represent a plurality of triphone phoneme models for which the number of trained frames having an equivalent effect on a phonemic context exceeds the threshold;
• and
• means for generating at least one shared phoneme model to represent a plurality of triphone phoneme models having the same center context.