A method of adapting a speech recognition system, wherein the method comprises steps of:</claim-text> a. obtaining an identification of a speaker; b. obtaining a sample of a speaker's speech during a first remote session; c. recognizing the speaker's ing the speech recognition system during the first remote session; d. modifying the speech recognition system by incorporating the sample into the speech recognition system thereby forming a speaker-specific modified speech recognition system; e. storing a representation of the speaker-specific modified speech recognition system in association with the identification of the speaker; and f. using the representation of the speaker-specific modified speech recognition system to recognize speech during a subsequent remote session with the speaker. view more

The method according to claim 1 further comprising a step of cumulatively modifying the speech recognition system according to speech samples obtained during one or more remote sessions with the speaker.

The method according to claim 4 wherein the representation of the speaker-specific modified acoustic model is a set of statistics which can be utilized to modify incoming acoustic speech.

A method of adapting a speech recognition system, wherein the method comprises steps of:</claim-text> a. obtaining an identification of a cluster of speakers; b. obtaining a sample of a speaker's speech during a first remote session; c. recognizing er's speech utilizing the speech recognition system during the first remote session; d. modifying the speech recognition system by incorporating the sample into the speech recognition system thereby forming a cluster-specific modified speech recognition system; e. storing a representation of the cluster-specific modified speech recognition system in association with the identification of a cluster of speakers wherein the speaker is a member of the cluster; and f. using the representation of the cluster-specific modified speech recognition system to recognize speech during a subsequent remote session with a member of the cluster of speakers. view more

The method according to claim 16 wherein the step of modifying the speech recognition system comprises a step of modifying an acoustic model thereby forming a cluster-specific modified acoustic model and wherein the step of storing a representation of the cluster-specific modified speech recognition system comprises a step of storing a representation of the cluster-specific modified acoustic model. view more

The method according to claim 16 wherein the step of modifying the speech recognition system is performed after termination of the first remote session.

A method of adapting a speech recognition system, wherein the method comprises steps of:</claim-text> a. obtaining an identification of each of a plurality of speakers during a corresponding first remote session with each speaker; b. obtaining a e of speech made by each of the plurality of speakers during a corresponding first remote session with each speaker; c. recognizing speech made by each speaker during the corresponding first remote session utilizing the speech recognition system configured to be speaker-independent; d. modifying the speech recognition system by individually incorporating the sample from each speaker into the speech recognition system thereby forming a speaker-specific modified speech recognition system corresponding to each speaker; e. storing a representation of the speaker-specific modified speech recognition system corresponding to each speaker in association with the identification of the corresponding speaker; and f. using the representation of the speaker-specific modified speech recognition system corresponding to a speaker to recognize speech during a subsequent remote session with the speaker. view more

The method according to claim 30 wherein the step of modifying the speech recognition system for each speaker is performed during the first remote session with the corresponding speaker.

A speech recognition system comprising:</claim-text> a. an interface coupled to receive a remote session from a speaker; and b. a processing system coupled to the interface to obtain an identification of the speaker and to recognize the speaker's erein the processing system is cumulatively modified by incorporating speech samples obtained during a plurality of remote sessions with the speaker into the speech recognition system, thereby forming a speaker-specific modified processing system associated with the identification of the speaker. view more

A method of adapting an acoustic model utilized for speech recognition, wherein the method comprises steps of:</claim-text> a. obtaining an identification of a speaker; b. obtaining a speech utterance from the speaker during a remote session; c. nizing the speaker's speech utilizing an acoustic model during the remote session; d. making a determination relative to the speech utterance; and e. only when indicated by the determination, performing steps of: i. modifying the acoustic model by incorporating the speech utterance into the acoustic model thereby forming a speaker-specific modified acoustic model; and ii. storing a representation of the speaker-specific modified acoustic model in association with the identification of the speaker. view more