89 results about "Phonetic transcription" patented technology

An adaptive speech recognition system is provided including an input for receiving a signal derived from a spoken utterance indicative of a certain vocabulary item, a speech recognition dictionary, a speech recognition unit and an adaptation module. The speech recognition dictionary has a plurality of vocabulary items each being associated to a respective dictionary transcription group. The speech recognition unit is in an operative relationship with the speech recognition dictionary and selects a certain vocabulary item from the speech recognition dictionary as being a likely match to the signal received at the input. The results of the speech recognition process are provided to the adaptation module. The adaptation module includes a transcriptions bank having a plurality of orthographic groups, each including a plurality of transcriptions associated with a common vocabulary item. A transcription selector module in the adaptation module retrieves a given orthographic group from the transcriptions bank on a basis of the vocabulary item recognized by the speech recognition unit. The transcription selector module processes the given orthographic group on the basis of the signal received at the input to select a certain transcription from the transcriptions bank. The adaptation module then modifies a dictionary transcription group corresponding to the vocabulary item selected as being a likely match to the signal received at the input on the basis of the selected certain transcription.

A method for building a language model, a speech recognition method and an electronic apparatus are provided. The speech recognition method includes the following steps. Phonetic transcriptions of a speech signal are obtained from an acoustic model. Phonetic spellings matching the phonetic transcriptions are obtained according to the phonetic transcriptions and a syllable acoustic lexicon. According to the phonetic spellings, a plurality of text sequences and a plurality of text sequence probabilities are obtained from a language model. Each phonetic spelling is matched to a candidate sentence table; a word probability of each phonetic spelling matching a word in a sentence of the sentence table are obtained; and the word probabilities of the phonetic spellings are calculated so as to obtain the text sequence probabilities. The text sequence corresponding to a largest one of the sequence probabilities is selected as a recognition result of the speech signal.

A method and a device (1) for automatically generating vocabulary entry from input acoustic data (3), comprising a vocabulary entry type-specific acoustic phonetic transcription module (2; T) and a classifier module (6; 6′) for the classification of vocabulary entry types on the basis of the phonetic structure, wherein the classification of vocabulary entries is carried out in accordance with a number of predetermined types; and vocabulary entry type-specific phoneme-to-grapheme conversion means (28), to derive the respective vocabulary entries comprising a pair of a phonetic transcription and its grapheme form.

The invention discloses a streaming phonetic transcription system based on a self-attention mechanism. The streaming phonetic transcription system based on the self-attention mechanism comprises a feature front-end processing module, a self-attention audio coding network module, a self-attention prediction network module and a united network module. The feature front-end processing module is usedfor receiving an input acoustic feature and converting into a vector with specific dimensionality; the self-attention audio coding network module is connected with the feature front-end processing module and is used for receiving the processed acoustic feature and obtaining an coded acoustic state vector; the self-attention prediction network module is used for generating a language state vector according to an input prediction mark of the last moment; and the united network module is connected with the self-attention audio coding network module and the self-attention prediction network module, and is used for combining with an acoustic state and a language state and calculating the probability of a new prediction mark. The invention provides a streaming feedforward voice coder based on the self-attention mechanism, so that the calculation efficiency and the precision of a traditional voice coder are improved.

The invention provides a speech recognition result error correction method and device. The speech recognition result error correction method comprises the steps of carrying out phonetic notation on a speech recognition result to be corrected so as to acquire phonetic transcription corresponding to the speech recognition result; acquiring candidate texts according to the phonetic transcription, and determining an optimal candidate text in the candidate texts; judging whether the optimal candidate text meets a preset condition or not; if the optimal candidate text meets the preset condition, determining the optimal candidate text as a correction result of the speech recognition result to be corrected. The speech recognition result error correction method can improve the accuracy of the correction result.

A system and method for generating synthetic speech, which operates in a computer implemented Text-To-Speech system. The system comprises at least a speaker database that has been previously created from user recordings, a Front-End system to receive an input text and a Text-To-Speech engine. The Front-End system generates multiple phonetic transcriptions for each word of the input text, and the TTS engine uses a cost function to select which phonetic transcription is the more appropriate for searching the speech segments within the speaker database to be concatenated and synthesized.

Methods for inputting characters of a symbol-based written language for encoding in a computerised system including a touch-sensitive surface, some methods comprising no more than four input steps. Movements of an object over the surface of the touch-sensitive surface, maintaining continuous contact therewith, defines a unique input path for each character. From a start position, an initial component in the input path selects a part of an alphabetical phonetic transcription associated with the character to be encoded. Groups of letters are displayed when initial contact is made, and, selection of one group displays individual letters within that group for selection. Once a selection has been made using the object, further components related to the character are displayed on the touch-sensitive surface at each step of the input path in accordance with previous selections, and if there is no ambiguity, removal of the object from the touch-sensitive surface encodes the character.

The invention relates to an intelligent mobile platform Pinyin (phonetic transcriptions of Chinese characters) input method based on a language models. The Pinyin input method comprises the following steps of: firstly, training a Pinyin text to obtain a language model based on letters and a language model based on Pinyin; secondly, decoding an input Pinyin string by using an HMM (Hidden Markov Model) decoding method; and thirdly, predicting a next input step and giving out an input promotion: firstly, carrying out prediction according to the language model based on the letters, and acquiring all reasonably-input letters which can occur behind a single Pinyin letter and the occurring probabilities of the reasonably-input letters; then, carrying out the prediction according to the language model based on the Pinyin, and acquiring all reasonably-input letters which can occur behind all possible Pinyin prefixes and the occurring probabilities of the reasonably-input letters; and finally, acquiring all next possible reasonably-input letters and the occurring probabilities of the possible reasonably-input letters by comprehensively considering information of the last two steps, comparing the probabilities, realizing the input prediction according to comparing results, and carrying out the input promotion. According to the intelligent mobile platform Pinyin input method based on the language models, the accuracy rate and the fluency of the input of a user are improved, and the input efficiency is greatly improved.

A system and method for searching for an element in speech related documents may include transcribing a set of speech recordings to a set of phoneme strings and including the phoneme strings in a set of phonetic transcriptions. A system and method may reverse-index the phonetic transcriptions according to one or more phonemes such that the one or more phonemes can be used as a search key for searching the phoneme in the phonetic transcriptions. A system and method may transcribe a textual search term into a set of search phoneme strings and use the set of search phoneme strings to search for an element in the set of phonetic transcriptions.

Technology related to improving synthesis of foreign regional nouns using personalized and culturally correct phonetic transcription is described. The technology includes systems and methods for generating personalized speech by receiving an input, the input including textual data, identifying a regional noun in the textual data, and determining a user accent classification based on a context of the input. The method may further include determining a personalized phonetic transcription of the regional noun corresponding to the user accent classification and using a phonetic inventory stored in a database and outputting the personalized phonetic transcription.

The invention relates to a word segmentation phonetic transcription and ligature writing method and device based on an SC grammar and belongs to the technical field of computer translation in computer science. Firstly, based on a word segmentation ambiguity rule of the SC grammar, an ambiguity segmentation rule library is built by means of abutment constraint conditions in natural language, and illegal segmentation is eliminated so that the word segmentation precision can be improved; secondly, based on a word segmentation ligature writing rule library of the SC grammar and a ligature writing corpora statistical library, the ligature writing corpora statistical library is used for performing ligature writing on ligature writing knowledge which cannot be presented as rules; finally, based on a dictionary library of the SC grammar, a dictionary is used for performing maximum matching to perform word segmentation, the word segmentation ambiguity rule is called for fields where ambiguity happens so that a correct segmentation result can be acquired, and the context of a word is analyzed so that correct part-of-speech tagging and phonetic transcription can be acquired. Compared with the prior art, word segmentation accuracy is improved, and the word segmentation ambiguity rule library, a combined ambiguity word library, the ligature writing rule library, the dictionary library and the ligature writing corpora statistical library are easy to expand and maintain.

A computer-based system generates alternative phonetic transcriptions for a target word or phrase corresponding to specific phonological processes that replace individual phonemes or clusters of two or more phonemes with replacement phonemes. The system compares a user's speech with a list of possible transcriptions that includes the base (i.e., correct) transcription of the test target as well as the different alternative transcriptions, to identify the transcription that best matches the user's. In a speech therapy application, the system identifies the phonological process(es), if any, associated with the user's speech and generates statistics over multiple test targets that can be used to diagnose the user's specific phonological disorders. The system can also be implemented in other contexts such as foreign language instruction and automated attendant applications to cover a wide variety and range of accents and/or phonological disorders.

A speech recognition method and an electronic apparatus are provided. The speech recognition method includes the following steps. A plurality of phonetic transcriptions of a speech signal is obtained according to an acoustic model. A phonetic spelling and intonation information matched to the phonetic transcriptions are obtained according to a phonetic transcription sequence and a syllable acoustic lexicon of the invention. According to the phonetic spellings and the intonation information, a plurality of phonetic spelling sequences and a plurality of phonetic spelling sequence probabilities are obtained from a language model. The phonetic spelling sequence corresponding to a largest one among the phonetic spelling sequence probabilities is selected as a recognition result of the speech signal.

The assignment of phonemes to graphemes producing them in a lexicon having words (grapheme sequences) and their associated phonetic transcription (phoneme sequences) for the preparation of patterns for training neural networks for the purpose of grapheme-phoneme conversion is carried out with the aid of a variant of dynamic programming which is known as dynamic time warping (DTW).

A method for training acoustic models for a new target language is provided using a phonetic table, which characterizes the phones, used in one or more reference language(s) with respect to their articulatory properties; a phonetic table, which characterizes the phones used in the target language with respect to their articulatory properties; a set of trained monophones for the reference language(s); and a database of sentences in the target language and its phonetic transcription. With these inputs, the new method completely and automatically takes care of the steps of monophone seeding and triphone clustering and machine intensive training steps involved in triphone acoustic training.

The embodiment of the invention discloses a plurilingual voice decoding diagram establishment method, a device, a server and a medium and relates to the technical field of voice recognition. The method comprises the following steps: marking main language words and secondary language words in a sample corpus bank with phonetic symbols so as to obtain pronunciation phonemes of the main language words and the secondary language words; according to sample voice associated with sample corpora in the sample corpus bank, confirming acoustic features of the main language words and the secondary language words; according to the main language words and the secondary language words in the sample corpora in the sample corpus bank, and the pronunciation phonemes and the acoustic features of the main language words and the secondary language words, confirming decoding diagrams of plurilingual recognition. According to the embodiment of the invention, the pronunciation phonemes of the main language words and the secondary language words are obtained according to the sample corpus bank, furthermore acoustic features associated with the main language words and the secondary language words are confirmed, the decoding diagrams of plurilingual recognition are finally obtained, and the requirement of voice recognition for plurilingual mixed reading crowds can be met.

A teaching machine generates a list of likely completions of an incomplete typed word, based upon previous keyboard input. This may include not only the incompletely typed word, but a number of completely typed, preceding words, in order have the word completion based upon context. The incompletely typed word is then subjected to a phonetic transcription, or other tests based upon knowledge by the system of the user, to further narrow the prediction list.

A process provides for searching through a written text in response to a spoken question comprising a plurality of words. The first step in the process is to transcribe the written text into a first sequence of phonetic units. Then, a spoken question is segmented into a second sequence of phonetic units. The search is conducted through the written text for an occurrence of the spoken question. The search comprises aligning the first and second sequences of phonetic units.

The method includes the following steps: 1) initial consonant, vowel, pronunciation tone and writing stroke of Chinese phonetic transcription are used as coding element separately and the table for them is set up according to dual-spelling of Chinese phonetic transcription; 2) coding index list is defined by using the phonetic transcription as index object; 3) phonetic transcription address index list is defined and 4) the system is firstly to search storing address of Chinese character according to dual-spelling inputted by the user, then all Chinese character with the same tone and stroke as inputted are searched out in relevant offset address and they are displayed on selection region, finally determinant matrix is applied for picking up Chinese character to be inputted.

The method described herein provides for comparing a similarity between various phonetic transcriptions of a specific foreign word on the basis of an English pronunciation similarity comparison algorithm.

The present invention provides multilanguage input. The input of stroke, phonetic transcription and combination are included in Chinese mode input/ intelligent English input, digital input and English-Chinese free switching input are also included. The functions of coding fault-tolerant, intelligent association are ready for each kind of input-mode. The coding principle is according to selfrule of various character so the boundary is more friendly with easy and quick operation.

An English spelling and reading method includes creating 144 colored phonetic symbols, using colored phonetic symbols to label English pronunciation, utilizing five formed module to divided key sounds, utilizing colored phonetic symbols to spell and to read key sounds then spelling and reading out pronunciation of English word continuously. The English spelling-reading machine used for realizing said method is also disclosed.