Avaya Technology LLC

Apple Inc.

Loral Aerospace Corp.

Octel Communications Corporation

Apple Computers Inc.

Kamil A Grajski

Peter Hunt

Leonardo Rub

Susannah Albright

Yen-Lu Chow

Kai-Fu Lee

Carson Chen

Garry Chinn

An apparatus for improving productivity of human reviewers of transcribed documents generated by media conversion systems includes a server/client network of computers, memories and file systems. The server receives and stores voice files created by users of the system. The server is configured for coupling to a speech-to-text media conversion system to receive converted text files of the audio voice files. The server analyzes the converted text files and routes the converted files to the appropriate reviewers according to an adaptive algorithm. The converted files are displayed on the assigned reviewer's screen at the reviewer's workstation. To aid the reviewer in pinpointing potential errors, the workstation displays different segments of the converted files in different colors to reflect different confidence levels of transcription accuracy. Portions of the original voice message that correspond to the potential errors are played back for the reviewer. The reviewers' workstations also perform productivity enhancing functions such as spelling and grammar checking. After the reviewer has made all the necessary corrections, the reviewed files are transmitted back to the server to be stored and accessed by the users. A user database in the server is also updated to store recurrent user-specific errors corrected by the reviewer. A language analysis system is also disposed to adaptively correct user-specific errors in future reviews according to the information in the user database.

Method of and apparatus for improving productivity of human reviewers of automatically transcribed documents generated by media conversion systems

A system and method for speech-responsive voice messaging, in which a Speech-Responsive Voice Messaging System (SRVMS) preferably provides a hierarchically-simple speech user interface (UI) that enables subscribers to use speech to specify commands such as mailboxes, passwords, and digits. The SRVMS generates and evaluates candidate results. The SRVMS invokes a speech UI navigation operation or a voice messaging operation according to the outcome of the evaluation of the candidate results. In the preferred embodiment, the SRVMS determines whether the candidate results are good, questionable, or bad; and whether two or more candidate results are ambiguous due to a likelihood that each such result could be a valid command. If the candidate results are questionable or ambiguous, an ambiguity resolution UI prompts the subscriber to confirm whether the best candidate result is what the subscriber intended. In response to repeated speech recognition failures, the SRVMS transfers the subscriber to a Dual Tone Multi Frequency (DTMF) UI. Transfer to the DTMF UI is also performed in response to detection of predetermined DTMF signals issued by the subscriber while the speech UI is in context. The SRVMS provides a logging unit and a reporting unit which operate in parallel with the speech UI, in a manner that is transparent to subscribers. The logging unit directs the selective logging of subscriber utterances, and the reporting unit selectively generates and maintains system performance statistics on multiple detail levels.

Speech-responsive voice messaging system and method

A system and method for processing stroke-based handwriting data for the purposes of automatically scoring and clustering the handwritten data to form letter prototypes. The present invention includes a method for processing digitized stroke-based handwriting data of known character strings, where each of the character strings is represented by a plurality of mathematical feature vectors. In this method, each one of the plurality of feature vectors is labelled as corresponding to a particular character in the character strings. A trajectory is then formed for each one of the plurality of feature vectors labelled as corresponding to a particular character. After the trajectories are formed, a distance value is calculated for each pair of trajectories corresponding to the particular character using dynamic time warping method. The trajectories which are within a sufficiently small distance of each other are grouped to form a plurality of clusters. The clusters are used to define handwriting prototypes which identify subcategories of the character.

Automatic method for scoring and clustering prototypes of handwritten

A system for automatic subcharacter unit and lexicon generation for handwriting recognition comprises a processing unit, a handwriting input device, and a memory wherein a segmentation unit, a subcharacter generation unit, a lexicon unit, and a modeling unit reside. The segmentation unit generates feature vectors corresponding to sample characters. The subcharacter generation unit clusters feature vectors and assigns each feature vector associated with a given cluster an identical label. The lexicon unit constructs a lexical graph for each character in a character set. The modeling unit generates a Hidden Markov Model for each set of identically-labeled feature vectors. After a first set of lexical graphs and Hidden Markov Models have been created, the subcharacter generation unit determines for each feature vector which Hidden Markov Model produces a highest likelihood value. The subcharacter generation unit relabels each feature vector according to the highest likelihood value, after which the lexicon unit and the modeling unit generate a new set of lexical graphs and a new set of Hidden Markov models, respectively. The feature vector relabeling, lexicon generation, and Hidden Markov Model generation are performed iteratively until a convergence criterion is met. The final set of Hidden Markov Model model parameters provide a set of subcharacter units for handwriting recognition, where the subcharacter units are derived from information inherent in the sample characters themselves.

System and method for automatic subcharacter unit and lexicon generation

A handwriting signal processing front-end method and apparatus for a handwriting training and recognition system which includes non-uniform segmentation and feature extraction in combination with multiple vector quantization. In a training phase, digitized handwriting samples are partitioned into segments of unequal length. Features are extracted from the segments and are grouped to form feature vectors for each segment. Groups of adjacent from feature vectors are then combined to form input frames. Feature-specific vectors are formed by grouping features of the same type from each of the feature vectors within a frame. Multiple vector quantization is then performed on each feature-specific vector to statistically model the distributions of the vectors for each feature by identifying clusters of the vectors and determining the mean locations of the vectors in the clusters. Each mean location is represented by a codebook symbol and this information is stored in a codebook for each feature. These codebooks are then used to train a recognition system. In the testing phase, where the recognition system is to identify handwriting, digitized test handwriting is first processed as in the training phase to generate feature-specific vectors from input frames. Multiple vector quantization is then performed on each feature-specific vector to represent the feature-specific vector using the codebook symbols that were generated for that feature during training. The resulting series of codebook symbols effects a reduced representation of the sampled handwriting data and is used for subsequent handwriting recognition.

Handwriting signal processing front-end for handwriting recognizers

An electronic detection system for detecting vibration patterns from physical events, such as engine knocking or articulated speech, senses a vibration source with a wideband transducer to provide a sensed signal, then converts this signal to energy amplitude and spectral frequency form to extract detectable features of the vibration pattern. Events are detected when a controller determines that the sensed energy and spectral data exceed adaptively-predetermined energy and spectral thresholds, preferably for a duration exceeding predefined time windows.

Growing community of inventors

San Jose, CA, United States of America

Kamil A Grajski