Abstract: An automated speech recognition (ASR) transcript of at least a portion of a media content is obtained from an ASR tool. Suggested words are received for corrected words of the ASR transcript of the media content. Features are obtained using at least the suggested words or the corrected words. The features include features relating to sound similarities between the suggested words and the corrected words. The features are input into a machine learning (ML) model to obtain a determination regarding a validity of the suggested words. Responsive to the suggested words constituting a valid suggestion, the suggested words are incorporated into the ASR transcript. At least a portion of the ASR transcript is transmitted to a user device in conjunction with at least a portion of the media content.

Abstract: Described herein is a system for encoding audio watermarks with frequency extensions to enable enhanced watermark detection. An extended audio watermark may include an existing audio watermark and a duplicate audio watermark, enabling backwards compatibility with existing watermark detection while also enabling enhanced watermark detection with increased accuracy. For example, embedding the extended audio watermark enables (i) limited devices to perform watermark detection to detect the existing audio watermark, and (ii) improved devices to perform enhanced watermark detection to detect the extended audio watermark. As the extended audio watermark includes redundancy in the form of duplicate audio watermark(s), an accuracy of performing enhanced watermark detection is increased relative to detecting the existing audio watermark alone.

Abstract: Methods and systems are described for generating dynamic conversational responses sensitive to different emotional contexts using machine learning models. The dynamic conversational responses may be generated in real time and reflect the likely emotional context by detecting socially close entities and events in user input.

Abstract: A question-and-answer system providing an appropriate answer for a vehicle-related FAQ by using deep learning, and corresponding method are provided. The question-and-answer system includes: a memory that stores a plurality of representative questions to match a plurality of answers corresponding to the plurality of representative questions, respectively; a learning module configured to output a representative question corresponding to an input sentence from among the stored plurality of representative questions; and an output module configured to search the memory or an answer that matches the output representative question and output the found answer.

Abstract: Various embodiments provide for summarization of an interaction, conversation, encounter, and/or the like in at least an abstractive manner. In one example embodiment, a method is provided. The method includes generating, using an encoder-decoder machine learning model, a party-agnostic representation data object for each utterance data object. The method further includes generating an attention graph data object to represent semantic and party-wise relationships between a plurality of utterance data objects. The method further includes modifying, using the attention graph data object, the party-agnostic representation data object for each utterance data object to form a party-wise representation data object for each utterance data object. The method further includes selecting a subset of party-wise representation data objects for each of a plurality of parties.

Abstract: A speech-processing system may determine potential endpoints in a user's speech. Such endpoint prediction may include determining a potential endpoint in a stream of audio data, and may additionally including determining an endpoint score representing a likelihood that the potential endpoint represents an end of speech representing a complete user input. When the potential endpoint has been determined, the system may publish a transcript of speech that preceded the potential endpoint, and send it to downstream components. The system may continue to transcribe audio data and determine additional potential endpoints while the downstream components process the transcript. The downstream components may determine whether the transcript is complete; e.g., represents the entirety of the user input. Final endpoint determinations may be made based on the results of the downstream processing including automatic speech recognition, natural language understanding, etc.

Abstract: A computing system generates a plurality of training data sets for generating the NLP model. The computing system trains a teacher network to extract and classify tokens from a document. The training includes a pre-training stage where the teacher network is trained to classify generic data in the plurality of training data sets and a fine-tuning stage where the teacher network is trained to classify targeted data in the plurality of training data sets. The computing system trains a student network to extract and classify tokens from a document by distilling knowledge learned by the teacher network during the fine-tuning stage from the teacher network to the student network. The computing system outputs the NLP model based on the training. The computing system causes the NLP model to be deployed in a remote computing environment.

Abstract: The present technology relates to an information processing device and method and a program that make it possible to reduce the total number of objects while the influence on the sound quality is suppressed. The information processing device includes a pass-through object selection unit configured to acquire data of L objects and select, from the L objects, M pass-through objects whose data is to be outputted as it is, and an object generation unit configured to generate, on the basis of the data of multiple non-pass-through objects that are not the pass-through objects among the L objects, the data of N new objects, N being smaller than (L?M). The present technology can be applied to an information processing device.

Abstract: A translation method, an electronic device and a storage medium, which relate to the field of artificial intelligence technologies, such as machine learning technologies, information processing technologies, are disclosed. An implementation includes: acquiring an intermediate translation result generated by each of multiple pre-trained translation models for a to-be-translated specified sentence in a same iteration of a translation process, so as to obtain multiple intermediate translation results; acquiring a co-occurrence word based on the multiple intermediate translation results; and acquiring a target translation result of the specified sentence based on the co-occurrence word.

Abstract: The present disclosure provides a method of recognizing speech offline, electronic device, and a storage medium, relating to a field of artificial intelligence such as speech recognition, natural language processing, and deep learning. The method may include: decoding speech data to be recognized into a syllable recognition result; transforming the syllable recognition result into a corresponding text as a speech recognition result of the speech data.

Abstract: Techniques for selecting a skill to execute in response to a natural language input are described. A system may receive a natural language input, determine profile data associated with the natural language input, and determine the profile data indicates a locale and at least first language and second languages. The system determines first and second sets of skills corresponding to the locale/first language and locale/second language, respectively. The system determines a first group of skill candidates corresponding to a portion of the first set of skills, and determines a second group of skill candidates corresponding to a portion of the second set of skills. The system performs ranking processing across the first and second groups of skills to determine a best skill for responding to the natural language input. Thus, in some situations, the skill invoked may not correspond to the same language represented in the natural language input.

Abstract: An example method includes, at an electronic device: receiving an indication of a notification; in accordance with receiving the indication of the notification: obtaining one or more data streams from one or more sensors; determining, based on the one or more data streams, whether a user associated with the electronic device is speaking; and in accordance with a determination that the user is not speaking: causing an output associated with the notification to be provided.

Abstract: Disclosed herein are system, method, and computer program product embodiments for similarity scoring of sentences, while restricting distances between tokenized pairs in the sentences. An embodiment operates by determining a similarity of tokens between a first sequence of tokens and a second sequence of tokens to generate token pairs, determining a distance of relative positioning of token pairs in the first tokenized sequence and the second tokenized sequence and generating a score value that indicates the degree to which the first sentence matches the second sentence based on restricting matches to a maximum value of the distance of relative positions of the token pairs in the first tokenized sequence and the second tokenized sequence.

Abstract: Disclosed embodiments may include a method that may include receiving a corpus of unlabeled text documents, generating, using the first machine learning model, a first classification of each unlabeled text document in the corpus of unlabeled text documents as positive or negative, defining, using the first machine learning model and based on the first classification, a first subset of the unlabeled text documents and a second subset of the unlabeled text documents, generating, using the second machine learning model, a second classification of each unlabeled text document in the first subset of the unlabeled text documents as positive or negative, generating, using the third machine learning model, a third classification of each unlabeled text document in the second subset of the unlabeled text documents as positive or negative, and modifying the first classification, based on the second classification and the third classification, to create a fourth classification.

Abstract: A word-sense disambiguation service may be performed to determine the semantic context of an ambiguous targeted word in an electronic data corpus. The word-sense disambiguation service may determine the semantic context of the words in the electronic data corpus by evaluating a main word and the context words surrounding the main word in a portion of text, then determine which context words are useful in defining the semantic context of the main word. The word-sense disambiguation service may then cluster the defining context words together and use the defining context words to train a word-embedding model to recognize the semantic context of an instance of the main word based on the proximity of the defining context words to the main word. A context-sensitive service may then receive input of a desired target word, then retrieve and display the various semantic contexts of the desired target word using the results gathered by the word-sense disambiguation service.

Abstract: Introduced here are computer programs and associated computer-implemented techniques for facilitating the creation of a master transcription (or simply “transcript”) that more accurately reflects underlying audio by comparing multiple independently generated transcripts. The master transcript may be used to record and/or produce various forms of media content, as further discussed below. Thus, the technology described herein may be used to facilitate editing of text content, audio content, or video content. These computer programs may be supported by a media production platform that is able to generate the interfaces through which individuals (also referred to as “users”) can create, edit, or view media content. For example, a computer program may be embodied as a word processor that allows individuals to edit voice-based audio content by editing a master transcript, and vice versa.

Abstract: One or more systems, devices, computer program products and/or computer-implemented methods of use provided herein relate to outputting an optimal decision policy base on informal knowledge input. A system can comprise a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory, wherein the computer executable components can comprise an analysis component that analyzes an input dataset comprising a constraint in a natural language form, and an augmentation component that generates an influence mapping comprising a constraint variable based on the constraint input. In an embodiment, an input dataset employed to support the influence mapping can comprise time-stamped tuple data comprising a state, an action and a reward. In an embodiment, an inference engine can generate an output policy in response to the constraint input and which output policy can be based on the constraint input and constraint variable.

Abstract: There is provided a learning device for learning a neural network used for search of external knowledge in order to increase search accuracy of external knowledge required for arithmetic processing. With an input sentence Q as an input, an external knowledge search unit 22 selects pieces of external knowledge based on similarity degrees between pieces of external knowledge included in an external knowledge database 2 and the input sentence Q, using a neural network, and causes the selected pieces of external knowledge to be search results R2. A processing unit 14 acquires a response sentence A to the input sentence Q by arithmetic processing with the input sentence Q and the selected pieces of external knowledge as an input. A consideration calculation unit 23 calculates a consideration v determined from an index indicating correctness of the response sentence A based on a true output T given to the input sentence Q in advance and an index indicating quality of the selected pieces of external knowledge.

Abstract: The subject technology groups received data in data blocks having a predetermined number of bytes. For each received data block, a compressed data block is written to an output buffer. The compressed data block includes a mask block having a same number of bits as the predetermined number, and a subsequent block. The mask block includes in a same order as bytes within the corresponding data block, a zero corresponding to a zero-byte within the data block, and a one corresponding to each non-zero byte within the data block. The subsequent block includes non-zero bytes within the corresponding data block in a same order as the non-zero bytes within the data block.

Abstract: Implementations of the present disclosure relate to text generation with a customizable style. In a method, a first natural language is received; the first natural language text is converted, via a text generation model, into a second natural language text that at least partly reflects the meaning of the first natural language text and has a style distinguishable from the first natural language text, the text generation model comprising a modifiable parameter; and in response to receiving a modification to the parameter, the first natural language text is converted, via the text generation model, into a third natural language text that at least partly reflects the meaning of the first natural language text and includes a style distinguishable from both the first natural language text and the second natural language text.