Samsung Electronics Co., Ltd.

In-jeong Choi

Jeong-su Kim

Sang-Bae Jeong

Nam-hoon Kim

Jae-Won Lee

Jae-Hoon Jeong

Sang-ryong Kim

Ick-Sang Han

Ick-sang Han

Su-yeon Yoon

An apparatus, method, and medium for dialogue speech recognition using topic domain detection are disclosed. An apparatus includes a forward search module performing a forward search in order to create a word lattice similar to a feature vector, which is extracted from an input voice signal, with reference to a global language model database, a pronunciation dictionary database and an acoustic model database, which have been previously established, a topic-domain-detection module detecting a topic domain by inferring a topic based on meanings of vocabularies contained in the word lattice using information of the word lattice created as a result of the forward search, and a backward-decoding module performing a backward decoding of the detected topic domain with reference to a specific topic domain language model database, which has been previously established, thereby outputting a speech recognition result for an input voice signal in text form. Accuracy and efficiency for a dialogue sentence are improved.

Apparatus, method, and medium for dialogue speech recognition using topic domain detection

Disclosed herein is a method and apparatus to recognize speech by measuring the confidence levels of respective frames. The method includes the operations of obtaining frequency features of a received speech signal for the respective frames having a predetermined length, calculating a keyword model-based likelihood and a filler model-based likelihood for each of the frame, calculating a confidence score based on the two types of likelihoods, and deciding whether the received speech signal corresponds to a keyword or a non-keyword based on the confidence scores. Also, the method includes the operation of transforming the confidence scores by applying transform functions of clusters, which include the confidence scores or are close to the confidence scores, to the confidence scores.

Method and apparatus for recognizing speech by measuring confidence levels of respective frames

An apparatus and method for recognizing voice. The apparatus includes a feature vector extraction unit dividing an input voice signal into predetermined unit regions, and extracting feature vectors corresponding to each of the unit regions; a predicted node extraction unit extracting a list of second nodes whose travels to a first node corresponding to the extracted feature vectors are predicted, with reference to a network of one or more nodes; a single waveform similarity calculation unit calculating degrees of single waveform similarity of the first node and the second nodes of the list by substituting the extracted feature vectors into single waveform probability distributions that constitute voice signals corresponding to the second nodes; a multiple waveform similarity calculation unit calculating degrees of multiple waveform similarity by substituting the extracted feature vectors into multiple waveform probability distributions that constitute single waveform probability distributions usable to calculate the degrees of single waveform similarity in a preset range; and an output unit outputting a function-performing signal corresponding to a multiple waveform probability distribution that enables calculation of a highest of the calculated degrees of multiple waveform similarity.

Apparatus and method for recognizing voice

A method and an apparatus for selecting a vocabulary closest to an input speech from among lexicons stored in memory, wherein a centroid lexicon representing lexicons belonging to a predetermined lexicon group is generated. Two lexicons, having a longest distance therebetween in the lexicon group, are selected using the centroid lexicon from the lexicon group, and a node indicating the lexicon group branches based on the two selected lexicons. A node having low group similarity is selected from among current terminal nodes, including branch nodes, and the above procedure is repeatedly performed on a lexicon group indicated by the selected node.

Speech recognition method and apparatus using lexicon group tree

A method and apparatus for discriminative estimation of parameters in a maximum a posteriori (MAP) speaker adaptation condition, and a voice recognition apparatus having the apparatus and a voice recognition method using the method are provided. The method for discriminative estimation of parameters in a maximum a posteriori (MAP) speaker adaptation condition, in which at least speaker-independent model parameters and prior density parameters, which are standards in recognizing a speaker's voice, are obtained as the result of model training after fetching training sets on a plurality of speakers from a training database, has the steps of (a) classifying adaptation data among training sets for respective speakers; (b) obtaining model parameters adapted from adaptation data on each speaker by using the initial values of the parameters; (c) searching a plurality of candidate hypotheses on each uttered sentence of training sets by using the adapted model parameters, and calculating gradients of speaker-independent model parameters by measuring the degree of errors on each training sentence; and (d) when training sets of all speakers are adapted, updating parameters, which were set at the initial stage, based on the calculated gradients.

Method and apparatus for discriminative estimation of parameters in maximum a posteriori (MAP) speaker adaptation condition and voice recognition method and apparatus including these

A continuous speech recognition method and system are provided. The continuous speech recognition method includes constructing a pronunciation dictionary database including at least one pronunciation representation for each word which is influenced by applying phonological rules, wherein the pronunciation representation for the coda of a first word or the pronunciation representation for the onset of a second word following the first word is additionally indexed with an identifier if it does not match the phonetic pronunciation of its spelling, forming inter-word phonetic information in matrix form by combination of a number of all probable phonetic pairs, each of which is basically comprised of the coda of a first word and the onset of a second word following the first word, wherein the coda of the first word or the onset of the second word is indexed with an identifier if they undergo phonological changes and performing speech recognition on feature vectors extracted from an input speech signal with reference to the pronunciation dictionary database and the inter-word phonetic information.

Growing community of inventors

Hwaseong-si, South Korea

In-jeong Choi