Nuance Communications, Inc.

Maximilian Bisani

Olivier Divay

William F Ganong, Iii

Paul Joseph Vozila

Bhuvana Ramabhadran

Stanley S Chen

Stephen Mingyu Chu

Ruhi Sarikaya

Jeffrey S Kobal

Yi Su

Alfred Dielmann

Methods and apparatus for speech recognition on user dictated words to generate a dictation and using a discriminative statistical model derived from a deterministic formatting grammar module and user formatted documents to extract features and estimate scores from the formatting graph. The processed dictation can be output as formatted text based on a formatting selection to provide an integrated stochastic and deterministic formatting of the dictation.

Methods and apparatus for joint stochastic and deterministic dictation formatting

A computer implemented method and system of formatting text output from a speech recognition system is provided. The method includes determining if a user correction to a text output from a speech recognition system can be accomplished by changing a formatting setting associated with the speech recognition system. The formatting setting is changed based on an inferential indication that the change to the formatting setting is acceptable to the user and/or an explicit confirmation from the user that the change to the formatting setting is acceptable.

Text formatter with intuitive customization

A language processing application employs a classing function optimized for the underlying production application context for which it is expected to process speech. A combination of class based and word based features generates a classing function optimized for a particular production application, meaning that a language model employing the classing function uses word classes having a high likelihood of accurately predicting word sequences encountered by a language model invoked by the production application. The classing function optimizes word classes by aligning the objective of word classing with the underlying language processing task to be performed by the production application. The classing function is optimized to correspond to usage in the production application context using class-based and word-based features by computing a likelihood of a word in an n-gram and a frequency of a word within a class of the n-gram.

Word classing for language modeling

Methods and apparatus for reducing security vulnerabilities in a client/server speech recognition system including one or more client computers and one or more server computers connected via a network. Decryption of sensitive information, such as medical dictation information, is performed on designated servers to limit the attack surface of unencrypted data. Management of encryption and decryption keys to restrict the storage and/or use of decryption keys on the server side of the client/server speech recognition system, while maintaining encrypted data on the server side is also described.

Securing speech recognition data

An input sequence of unstructured speech recognition text is transformed into output structured document text. A probabilistic word substitution model is provided which establishes association probabilities indicative of target structured document text correlating with source unstructured speech recognition text. The input sequence of unstructured speech recognition text is looked up in the word substitution model to determine likelihoods of the represented structured document text corresponding to the text in the input sequence. Then, a most likely sequence of structured document text is generated as an output.

Growing community of inventors

Limbourg, Belgium

Maximilian Bisani