International Business Machines Corporation

Nuance Communications, Inc.

The Ohio State University

Ohio State Innovation Foundation

Hong-Kwang Jeff Kuo

Samuel Jonathan Thomas

Brian E D Kingsbury

Yuqing Gao

Lidia Mangu

George Andrei Saon

Vaibhava Goel

Bowen Zhou

Cheng Wu

Michael Alan Picheny

David M Lubensky

Liang Gu

Kartik Audhkhasi

Sabine V Deligne

Hagen Soltau

Zoltan Tueske

Jatin Ganhotra

Andrzej Sakrajda

Yinghui Huang

Eric Fosler-Lussier

Vishal Sunder

Gakuto Kurata

David Nahamoo

Sachindra Joshi

Ramesh Ambat Gopinath

Ea-Ee Jan

Sreeram Viswanath Balakrishnan

Stanley S Chen

Benoit Maison

Geoffrey G Zweig

Yuging Gao

Scott Elliot Axelrod

Mohamed A Afify

Systems, computer-implemented methods, and computer program products to facilitate fine-grained textual knowledge transfer to improve speech recognition and understanding are provided. According to an embodiment, a system can comprise a processor that executes components stored in memory. The computer executable components comprise deriving component that can derive one or more speech-based embeddings from an utterance via a speech encoder. The computer executable components can comprise a cross-attention component that can align, at a token level, one or more large language model (LLM) based sentence embeddings with the one or more speech-based embeddings. The computer executable components can comprise a loss component that can combine an alignment loss and an automatic speech recognition (ASR) loss.

Textual knowledge transfer for improved speech recognition and understanding

Features of two or more single speaker utterances are concatenated together and corresponding labels of the two or more single speaker utterances are concatenated together. Single speaker acoustic embeddings for each of the single speaker utterances of the concatenated single speaker utterances are generated using a single speaker teacher encoder network. An enhanced model is trained on the concatenated single speaker utterances using a classification loss Land a representation similarity loss L, the representation similarity loss Ldefined to influence an embedding derived from the concatenated single speaker utterances, the influence being based on the single speaker acoustic embeddings derived from the single speaker teacher encoder network.

Multi-speaker data augmentation for improved end-to-end automatic speech recognition

Audio signals representing a current utterance in a conversation and a dialog history including at least information associated with past utterances corresponding to the current utterance in the conversation can be received. The dialog history can be encoded into an embedding. A spoken language understanding neural network model can be trained to perform a spoken language understanding task based on input features including at least speech features associated with the received audio signals and the embedding. An encoder can also be trained to encode a given dialog history into an embedding. The spoken language understanding task can include predicting a dialog action of an utterance. The spoken language understanding task can include predicting a dialog intent or overall topic of the conversation.

Integrating dialog history into end-to-end spoken language understanding systems

Systems, computer-implemented methods, and computer program products to facilitate end to end integration of dialogue history for spoken language understanding are provided. According to an embodiment, a system can comprise a processor that executes components stored in memory. The computer executable components comprise a conversation component that encodes speech-based content of an utterance and text-based content of the utterance into a uniform representation.

End-to-end integration of dialog history for spoken language understanding

Training data can be received, which can include pairs of speech and meaning representation associated with the speech as ground truth data. The meaning representation includes at least semantic entities associated with the speech, where the spoken order of the semantic entities is unknown. The semantic entities of the meaning representation in the training data can be reordered into spoken order of the associated speech using an alignment technique. A spoken language understanding machine learning model can be trained using the pairs of speech and meaning representation having the reordered semantic entities. The meaning representation, e.g., semantic entities, in the received training data can be perturbed to create random order sequence variations of the semantic entities associated with speech. Perturbed meaning representation with associated speech can augment the training data.

Training end-to-end spoken language understanding systems with unordered entities

A method and system of training a spoken language understanding (SLU) model includes receiving natural language training data comprising (i) one or more speech recording, and (ii) a set of semantic entities and/or intents for each corresponding speech recording. For each speech recording, one or more entity labels and corresponding values, and one or more intent labels are extracted from the corresponding semantic entities and/or overall intent. A spoken language understanding (SLU) model is trained based upon the one or more entity labels and corresponding values, and one or more intent labels of the corresponding speech recordings without a need for a transcript of the corresponding speech recording.

End-to-end spoken language understanding without full transcripts

A processor-implemented method trains an automatic speech recognition system using speech data and text data. A computer device receives speech data, and generates a spectrogram based on the speech data. The computing device receives text data associated with an entire corpus of text data, and generates a textogram based upon the text data. The computing device trains an automatic speech recognition system using the spectrogram and the textogram.

Integrating text inputs for training and adapting neural network transducer ASR models

Embodiments of the present invention provide computer implemented methods, computer program products and computer systems. For example, embodiments of the present invention can access one or more intents and associated entities from limited amount of speech to text training data in a single language. Embodiments of the present invention can locate speech to text training data in one or more other languages using the accessed one or more intents and associated entities to locate speech to text training data in the one or more other languages different than the single language. Embodiments of the present invention can then train a neural network based on the limited amount of speech to text training data in the single language and the located speech to text training data in the one or more other languages.

Multilingual intent recognition

An illustrative embodiment includes a method for training an end-to-end (E2E) spoken language understanding (SLU) system. The method includes receiving a training corpus comprising a set of text classified using one or more sets of semantic labels but unpaired with speech and using the set of unpaired text to train the E2E SLU system to classify speech using at least one of the one or more sets of semantic labels. The method may include training a text-to-intent model using the set of unpaired text; and training a speech-to-intent model using the text-to-intent model. Alternatively or additionally, the method may include using a text-to-speech (TTS) system to generate synthetic speech from the unpaired text; and training the E2E SLU system using the synthetic speech.

Leveraging unpaired text data for training end-to-end spoken language understanding systems

A method, executed by a computer, includes receiving a channel recording corresponding to a conversation, receiving a transcription for the conversation, generating a conversation-specific language model for the conversation using the transcription, and conducting speech recognition on the channel recording using the conversation-specific language model to provide time boundaries and written language corresponding to utterances within the channel recording. The method further includes determining sentence or phrase boundaries for the transcription, aligning written language within the one or more transcriptions with the written language corresponding to the utterances with the channel recording to provide sentence or phrase boundaries for the channel recording, and training a speech recognizer according to the sentence or phrase boundaries for the transcription and the sentence or phrase boundaries for the channel recording. A computer system and computer program product corresponding to the method are also disclosed herein.

Acoustic model training

An automatic speech recognition system and a method performed by an automatic speech recognition system are provided. The method includes performing at least two passes of speech activity detection on an acoustic utterance uttered by a speaker. The at least two passes include an initial pass and a subsequent pass. The method further includes estimating at least one of feature statistics and transforms for acoustic feature extraction and acoustic modeling based on an output of an initial pass. The method further includes performing automatic speech recognition using an output of the subsequent pass while bypassing an output of the initial pass to recognize the acoustic utterance.

Multi-pass speech activity detection strategy to improve automatic speech recognition

Systems and methods for spoken term detection are provided. A method for spoken term detection, comprises receiving phone level out-of-vocabulary (OOV) keyword queries, converting the phone level OOV keyword queries to words, generating a confusion network (CN) based keyword searching (KWS) index, and using the CN based KWS index for both in-vocabulary (IV) keyword queries and the OOV keyword queries.

Method and system for efficient spoken term detection using confusion networks

Systems and methods for processing a query include determining a plurality of sets of match candidates for a query using a processor, each of the plurality of sets of match candidates being independently determined from a plurality of diverse word lattice generation components of different type. The plurality of sets of match candidates is merged by generating a first score for each match candidate to provide a merged set of match candidates. A second score is computed for each match candidate of the merged set based upon features of that match candidate. The first score and the second score are combined to provide a final set of match candidates as matches to the query.

Growing community of inventors

Pleasantville, NY, United States of America

Hong-Kwang Jeff Kuo