Nuance Communications, Inc.

Nortel Networks Corporation

Northern Telecom Limited

Serge Robillard

Real Tremblay

Jerome Tremblay

Vishwa N Gupta

Michael G Sabourin

Tagyoung Chung

Alina Andreevskaia

Jean-Francois Crespo

Waleed Fakhr

Jackson Liscombe

Peter R Stubley

Stephen Douglas Peters

Ding Liu

Réal Tremblay

Andre Gillet

Nadia Girolamo

Aspects described herein provide quality assurance checks for improving the construction of natural language understanding grammars. An annotation module may obtain a set of annotations for a set of text samples based, at least in part, on an ontology and a grammar. A quality assurance module may automatically perform one or more quality assurance checks on the set of annotations, the ontology, the grammar, or combinations thereof. The quality assurance module may generate a list of flagged annotations during performance of a quality assurance check. The list of flagged annotations may be presented at an annotation review interface displayed at a display device. One of the flagged annotations may be selected and presented at an annotation interface displayed at the display device. Responsive to presentation of the flagged annotation, the ontology, the grammar, the flagged annotation selected, or combinations thereof may be updated based on user input received.

Automated quality assurance checks for improving the construction of natural language understanding systems

Inferring a natural language grammar is based on providing natural language understanding (NLU) data with concept annotations according to an application ontology characterizing a relationship structure between application-related concepts for a given NLU application. An application grammar is then inferred from the concept annotations and the application ontology.

Ontology and annotation driven grammar inference

In FAQ based systems, associating questions with answers can be a time consuming task if performed manually. In one embodiment, a method of building a frequently-asked questions (FAQ) portal can include creating cluster labels. The labels can include predefined universal semantic labels and application-specific labels. The method can further include applying the cluster labels to clusters of queries within an FAQ application. The method can additionally include adjusting the application-specific labels to support combined and newly created clusters of queries based on application-specific queries within the FAQ application on an ongoing basis and reapplying the universal semantic labels and the adjusted application-specific labels to the combined and newly created clusters of queries. The method and system proposed herein allow for the automated clustering of queries and association with applicable answers, which leads to higher efficiencies for a faster response time for a user.

Method and apparatus for a frequently-asked questions portal workflow

The invention relates to a method and apparatus for training a multilingual speech model set. The multilingual speech model set generated is suitable for use by a speech recognition system for recognizing spoken utterances for at least two different languages. The invention allows using a single speech recognition unit with a single speech model set to perform speech recognition on utterances from two or more languages. The method and apparatus make use of a group of a group of acoustic sub-word units comprised of a first subgroup of acoustic sub-word units associated to a first language and a second subgroup of acoustic sub-word units associated to a second language where the first subgroup and the second subgroup share at least one common acoustic sub-word unit. The method and apparatus also make use of a plurality of letter to acoustic sub-word unit rules sets, each letter to acoustic sub-word unit rules set being associated to a different language. A set of untrained speech models is trained on the basis of a training set comprising speech tokens and their associated labels in combination with the group of acoustic sub-word units and the plurality of letter to acoustic sub-word unit rules sets. The invention also provides a computer readable storage medium comprising a program element for implementing the method for training a multilingual speech model set.

Method and apparatus for training a multilingual speech model set

A system and method for continuous speech recognition (CSR) is optimized to reduce processing time for connected word grammars bounded by semantically null words. The savings, which reduce processing time both during the forward and the backward passes of the search, as well as during rescoring, are achieved by performing only the minimal amount of computation required to produce an exact N-best list of semantically meaningful words (N-best list of salient words). This departs from the standard Spoken Language System modeling which any notion of meaning is handled by the Natural Language Understanding (NLU) component. By expanding the task of the recognizer component from a simple acoustic match to allow semantic information to be fed to the recognizer, significant processing time savings are achieved, and make it possible to run an increased number of speech recognition channels in parallel for improved performance, which may enhance users perception of value and quality of service.

Search optimization system and method for continuous speech recognition

Speech recognition systems and methods consistent with the present invention process input speech signals organized into a series of frames. The input speech signal is decimated to select K frames out of every L frames of the input speech signal according to a decimation rate K/L. A first set of model distances is then calculated for each of the K selected frames of the input speech signal, and a Hidden Markov Model (HMM) topology of a first set of models is reduced according to the decimation rate K/L. The system then selects a reduced set of model distances from the computed first set of model distances according to the reduced HMM topology and selects a first plurality of candidate choices for recognition according to the reduced set of model distances. A second set of model distances is computed, using a second set of models, for a second plurality of candidate choices, wherein the second plurality of candidate choices correspond to at least a subset of the first plurality of candidate choices. The second plurality of candidate choices are rescored using the second set of model distances, and a recognition result is selected from the second plurality of candidate choices according to the rescored second plurality of candidate choices.

Search and rescoring method for a speech recognition system

Systems and methods consistent with the present invention determine whether to accept one of a plurality of intermediate recognition results output by a speech recognition system as a final recognition result. The system first combines a plurality of speech rejection features into a feature function in which weights are assigned to each rejection feature in accordance with a recognition accuracy of each rejection feature. Feature values are then calculated for each of the rejection features using the plurality of intermediate recognition results. The system next computes the feature function according to the calculated feature values to determine a rejection decision value. Finally, one of the plurality of intermediate recognition results is accepted as the final recognition result according to the rejection decision value.

Speech recognition rejection method using generalized additive models

The invention relates to a method and apparatus for grouping orthographies in a speech recognition dictionary to reduce false rejection. In a typical speech recognition system, the process of speech recognition consists of scanning the vocabulary database by using a fast match algorithm to find the top N orthography groups. In a second pass the orthographies in the selected groups are re-scored using more precise likelihoods. The top orthographies in the top two groups are then processed by a rejection algorithm to find if they are sufficiently distinct from one another. In the affirmative the top choice candidate is accepted, otherwise the entire procedure is terminated. The novel method comprises the steps of grouping confusable orthographies together to reduce the possibility of false rejection.

Growing community of inventors

Lachine, Canada

Serge Robillard