International Business Machines Corporation

Nuance Communications, Inc.

Abhinav Sethy

Bhuvana Ramabhadran

Stanley S Chen

Kenneth W Church

Samuel Jonathan Thomas

Vaibhava Goel

Etienne Marcheret

Lidia Mangu

Laurence P Sansone

Masayuki A Suzuki

Gakuto Kurata

Ryuki Tachibana

Paul Joseph Vozila

Kartik Audhkhasi

Ebru Arisoy

Ron Hoory

Jonathan Mamou

Nathan M Bodenstab

Yinghui Huang

Dimitri Kanevsky

Symbol sequences are estimated using a computer-implemented method including detecting one or more candidates of a target symbol sequence from a speech-to-text data, extracting a related portion of each candidate from the speech-to-text data, detecting repetition of at least a partial sequence of each candidate within the related portion of the corresponding candidate, labeling the detected repetition with a repetition indication, and estimating whether each candidate is the target symbol sequence, using the corresponding related portion including the repetition indication of each of the candidates.

Symbol sequence estimation in speech

A method of combining data streams from fixed audio-visual sensors with data streams from personal mobile devices including, forming a communication link with at least one of one or more personal mobile devices; receiving at least one of an audio data stream and/or a video data stream from the at least one of the one or more personal mobile devices; determining the quality of the at least one of the audio data stream and/or the video data stream, wherein the audio data stream and/or the video data stream having a quality above a threshold quality is retained; and combining the retained audio data stream and/or the video data stream with the data streams from the fixed audio-visual sensors.

Combining installed audio-visual sensors with ad-hoc mobile audio-visual sensors for smart meeting rooms

A computer selects a test set of sentences from among sentences applied to train a whole sentence recurrent neural network language model to estimate the probability of likelihood of each whole sentence processed by natural language processing being correct. The computer generates imposter sentences from among the test set of sentences by substituting one word in each sentence of the test set of sentences. The computer generates, through the whole sentence recurrent neural network language model, a first score for each sentence of the test set of sentences and at least one additional score for each of the imposter sentences. The computer evaluates an accuracy of the natural language processing system in performing sequential classification tasks based on an accuracy value of the first score in reflecting a correct sentence and the at least one additional score in reflecting an incorrect sentence.

Implementing a whole sentence recurrent neural network language model for natural language processing

A whole sentence recurrent neural network (RNN) language model (LM) is provided for for estimating a probability of likelihood of each whole sentence processed by natural language processing being correct. A noise contrastive estimation sampler is applied against at least one entire sentence from a corpus of multiple sentences to generate at least one incorrect sentence. The whole sentence RNN LN is trained, using the at least one entire sentence from the corpus and the at least one incorrect sentence, to distinguish the at least one entire sentence as correct. The whole sentence recurrent neural network language model is applied to estimate the probability of likelihood of each whole sentence processed by natural language processing being correct.

A mechanism is provided in a data processing system for external word embedding neural network language models. The mechanism configures the data processing system with an external word embedding neural network language model that accepts as input a sequence of words and predicts a current word based on the sequence of words. The external word embedding neural network language model combines an external embedding matrix to a history word embedding matrix and a prediction word embedding matrix of the external word embedding neural network language model. The mechanism receives a sequence of input words by the data processing system. The mechanism applies a plurality of previous words in the sequence of input words as inputs to the external word embedding neural network language model. The external word embedding neural network language model generates a predicted current word based on the plurality of previous words. The mechanism processes a current word in the sequence of input words based on the predicted current word generated by the external word embedding neural network language model.

External word embedding neural network language models

Techniques for conversion of non-back-off language models for use in speech decoders. For example, an apparatus for conversion of non-back-off language models for use in speech decoders. For example, an apparatus is configured convert a non-back-off language model to a back-off language model. The converted back-off language model is pruned. The converted back-off language model is usable for decoding speech.

Conversion of non-back-off language models for efficient speech decoding

Some embodiments relate to using an unnormalized neural network language model in connection with a speech processing application. The techniques include obtaining a language segment sequence comprising one or more language segments in a vocabulary; accessing an unnormalized neural network language model having a normalizer node and an output layer comprising a plurality of output nodes, each of the plurality of output nodes associated with a respective language segment in the vocabulary; and determining, using the unnormalized neural network language model, a first likelihood that a first language segment in the vocabulary follows the language segment sequence.

Systems and methods for providing unnormalized language models

Techniques for conversion of non-back-off language models for use in speech decoders. For example, a method comprises the following step. A non-back-off language model is converted to a back-off language model. The converted back-off language model is pruned. The converted back-off language model is usable for decoding speech.

Techniques disclosed herein include systems and methods for voice-enabled searching. Techniques include a co-occurrence based approach to improve accuracy of the 1-best hypothesis for non-phrase voice queries, as well as for phrased voice queries. A co-occurrence model is used in addition to a statistical natural language model and acoustic model to recognize spoken queries, such as spoken queries for searching a search engine. Given an utterance and an associated list of automated speech recognition n-best hypotheses, the system rescores the different hypotheses using co-occurrence information. For each hypothesis, the system estimates a frequency of co-occurrence within web documents. Combined scores from a speech recognizer and a co-occurrence engine can be combined to select a best hypothesis with a lower word error rate.

Growing community of inventors

Chappaqua, NY, United States of America

Abhinav Sethy