Nippon Telegraph and Telephone Corporation

Mitsubishi Electric Corporation

Renesas Technology Corp.

Ebara Corporation

Yahoo Japan Corporation

17live Japan Inc.

Jfe Steel Corporation

Nikon-trimble Co., Ltd

Tomohiro Tanaka

Ryo Masumura

Takashi Nakamura

Koichi Takeda

Kenichiro Saito

Yuji Miki

Takanobu Oba

Ryuichi Kosuge

Naofumi Murata

Tohru Koyama

Akitoshi Matsui

Naoaki Yamashita

Masumi Nishijima

Ryosuke Chiyohara

Masayuki Mitsuzono

Yoichi Itoh

A tag estimation device capable of estimating, for an utterance made among several persons, a tag representing a result of analyzing the utterance is provided. The tag estimation device includes an utterance sequence information vector generation unit that adds a t-th utterance word feature vector and a t-th speaker vector to a (t-1)-th utterance sequence information vector uthat includes an utterance word feature vector that precedes the t-th utterance word feature vector and a speaker vector that precedes the t-th speaker vector to generate a t-th utterance sequence information vector u, where t is a natural number, and a tagging unit that determines a tag lthat represents a result of analyzing a t-th utterance from a model parameter set in advance and the t-th utterance sequence information vector u.

Tag estimation device, tag estimation method, and program

A terminal of a viewer who views a live-stream includes one or more processors, and memory storing one or more computer programs configured to be executed by the one or more processors. The one or more computer programs includes instructions for: causing transition from a live-stream screen to a post-stream screen when a distributor of the live-stream ends the live-stream, the live-stream screen showing a video image generated by a terminal of the distributor, the post-stream screen showing a predetermined image; and accepting an instruction to use a gift from the viewer on the post-stream screen.

Terminal and method

A voice recognition deviceincludes: a phonological awareness feature amount extraction unitthat transforms an acoustic feature amount sequence of input voice into a phonological awareness feature amount sequence for the languageusing a first model parameter group; a phonological awareness feature amount extraction unitthat transforms the acoustic feature amount sequence of the input voice into a phonological awareness feature amount sequence for the languageusing a second model parameter group; a phonological recognition unitthat generates a posterior probability sequence from the acoustic feature amount sequence of the input voice, the phonological awareness feature amount sequence for the language, and the phonological awareness feature amount sequence for the languageusing a third model parameter group; and a voice text transformation unitthat performs voice recognition based on the posterior probability sequence to output text of a voice recognition result.

Learning device, voice recognition device, learning method, voice recognition method, and program

Without dividing speech into a unit such as a word or a character, text corresponding to the speech is labeled. A speech distributed representation sequence converting unitconverts an acoustic feature sequence into a speech distributed representation. A symbol distributed representation converting unitconverts each symbol included in the symbol sequence corresponding to the acoustic feature sequence into a symbol distributed representation. A label estimation unitestimates a label corresponding to the symbol from the fixed-length vector of the symbol generated using the speech distributed representation, the symbol distributed representation, and fixed-length vectors of previous and next symbols.

Sequence labeling apparatus, sequence labeling method, and program

A keyword is extracted robustly despite a voice recognition result including an error. A model storage unitstores a keyword extraction model that accepts word vector representations of a plurality of words as an input and extracts and outputs a word vector representation of a word to be extracted as a keyword. A speech detection unitdetects a speech part from a voice signal. A voice recognition unitexecutes voice recognition on the speech part of the voice signal and outputs a confusion network which is a voice recognition result. A word vector representation generating unitgenerates a word vector representation including reliability of voice recognition with regard to each candidate word for each confusion set. A keyword extraction unitinputs the word vector representation of the candidate word to the keyword extraction model in descending order of the reliability and obtains the word vector representation of the keyword.

Keyword detection apparatus, keyword detection method, and program

A language model score calculation apparatus calculates a prediction probability of a word was a language model score of a language model based on a recurrent neural network. The language model score calculation apparatus includes a memory; and a processor configured to execute converting a word wthat is observed immediately before the word winto a word vector ϕ(w); converting a speaker label rcorresponding to the word wand a speaker label rcorresponding to the word winto a speaker vector ψ(r) and a speaker vector ψ(r), respectively; calculating a word history vector sby using the word vector ϕ(w), the speaker vector ψ(r), and a word history vector sthat is obtained when a prediction probability of the word wis calculated; and calculating a prediction probability of the word wby using the word history vector sand the speaker vector ψ(r).

Language model score calculating apparatus, learning apparatus, language model score calculating method, learning method and program

Provided is a speech recognition device capable of implementing end-to-end speech recognition considering a context. The speech recognition device includes a model parameter learning unit that learns a model parameter θ by using a word sequence of concern as an observation value and using a word sequence previous to the word sequence of concern, an acoustic feature value sequence corresponding to the word sequence of concern, and the model parameter θ as parameters to perform maximum likelihood estimation for a likelihood function of a probability that the observation value occurs under the parameters and an uttered speech recognition unit that repeats, in order of time sequence, processing of recognizing a word sequence to be recognized, the processing of recognizing the word sequence to be recognized being performed by using the word sequence to be recognized as an observation value and using an already recognized word sequence previous to the word sequence to be recognized, an acoustic feature value sequence corresponding to the word sequence to be recognized, and the learned model parameter θ as parameters and based on a maximum likelihood criterion for the likelihood function of the probability that the observation value occurs under the parameters.

Speech recognition device, speech recognition method, and program

A tag estimation device capable of estimating, for an utterance made among several persons, a tag representing a result of analyzing the utterance is provided. The tag estimation device includes an utterance sequence information vector generation unit that adds a t-th utterance word feature vector and a t-th speaker vector to a (t−1)-th utterance sequence information vector uthat includes an utterance word feature vector that precedes the t-th utterance word feature vector and a speaker vector that precedes the t-th speaker vector to generate a t-th utterance sequence information vector u, where t is a natural number, and a tagging unit that determines a tag lthat represents a result of analyzing a t-th utterance from a model parameter set in advance and the t-th utterance sequence information vector u.

The present invention improves the accuracy of language prediction. A history speech meta-information understanding unitobtains a history speech meta-information vector from a word string of a preceding speech using a meta-information understanding device. A history speech embedding unitconverts the word string of the preceding speech and a speaker label into a history speech embedding vector. A speech unit combination vector construction unitobtains a speech unit combination vector by combining the history speech meta-information vector and the history speech embedding vector. A speech sequence embedding vector calculation unitconverts a plurality of speech unit combination vectors obtained for the past speech sequences to a speech sequence embedding vector. A language model score calculation unitcalculates a language model score of a current speech from a word string of the current speech, a speaker label, and a speech sequence embedding vector.

Language model score calculation apparatus, language model generation apparatus, methods therefor, program, and recording medium

An air condition control device includes a calculating circuitry to calculate a control influence degree, intended for controlling air condition changing devices, at each of control points in a room space on the basis of positional information that indicates a position where each of the air condition changing devices is located, influence degree information that indicates an influence degree of an air condition exerted on each of locations by each of the air condition changing devices, and running condition information that indicates a running condition of each of the air condition changing devices. The control influence degree is an influence degree of an air condition exerted by the air condition changing devices. The air condition control device also includes a controlling circuit to control the air condition changing devices to increase the control influence degree at a control point, which is relatively low.

Air condition control device

A learning apparatus comprises a learning part that learns an error correction model by a set of a speech recognition result candidate and a correct text of speech recognition for given audio data, wherein the speech recognition result candidate includes a speech recognition result candidate which is different from the correct text, and the error correction model is a model that receives a word sequence of the speech recognition result candidate as input and outputs an error correction score indicating likelihood of the word sequence of the speech recognition result candidate in consideration of a speech recognition error.

Learning apparatus, speech recognition rank estimating apparatus, methods thereof, and program

A generating apparatus according to an embodiment of the present application includes an obtaining unit and a generating unit. The obtaining unit obtains intermediate images in an intermediate layer of a neural network that recognizes the quantity of objects contained in an image. The generating unit generates a combination image by combining together the intermediate images obtained by the obtaining unit. For example, the generating unit generates the combination image by combining together the intermediate images to which a weight has been applied on the basis of a change in the recognition of the quantity of objects corresponding to a process applied to each of the intermediate images.

Generating apparatus, generating method, and non-transitory computer readable storage medium

A substrate processing apparatus includes transport mechanismstoa sensorfor detecting whether a wafer W is held by the transport mechanismstoa sensor controllerfor controlling the sensora position information acquiring partfor acquiring position information of the substrate W in the transport passage, a detection result acquiring partfor acquiring a detection result from the sensorand a determination processorfor determining whether the detection result of the sensor and the position information are consistent with each other. When it is determined that the detection result of the sensoris inconsistent with the position information, the sensor controllermakes the sensorretry to detect whether the substrate W is held.

Substrate processing apparatus and substrate presence or absence checking method and program

A method of manufacturing a semiconductor device includes forming a first interlayer insulation film on a semiconductor substrate, depositing a first metal film on the first interlayer insulation film, depositing an antireflection film including a dielectric layer on an upper surface of the first metal film, patterning the first metal film and the antireflection film to form a lower electrode having the antireflection film on an upper surface thereof, forming a second interlayer insulation film on the antireflection film, forming first and second openings in a first region and in a second region in the second interlayer insulation film, respectively, removing a portion of the antireflection film where the second opening is formed, depositing a second metal film on the second interlayer insulation film, and removing the second metal film except in the first and second openings to form an upper electrode in the first opening, and a contact in the second opening.

Growing community of inventors

Tokyo, Japan

Tomohiro Tanaka