International Business Machines Corporation

Peter F Brown

Robert L Mercer

Stephen A Della Pietra

Vincent J Della Pietra

Frederick Jelinek

Lalit R Bahl

Jennifer Ceil Lai

John Cocke

Peter V DeSouza

Adam Lee Berger

Michael Alan Picheny

Mark Edward Epstein

Stanley S Chen

Peter V De Souza

Steven V De Gennaro

Peter V Desouza

John David Lafferty

Andrew S Kehler

David Nahomoo

Philip S Resnik

An apparatus for translating a series of source words in a first language to a series of target words in a second language. For an input series of source words, at least two target hypotheses, each comprising a series of target words, are generated. Each target word has a context comprising at least one other word in the target hypothesis. For each target hypothesis, a language model match score comprises an estimate of the probability of occurrence of the series of words in the target hypothesis. At least one alignment connecting each source word with at least one target word in the target hypothesis is identified. For each source word and each target hypothesis, a word match score comprises an estimate of the conditional probability of occurrence of the source word, given the target word in the target hypothesis which is connected to the source word and given the context in the target hypothesis of the target word which is connected to the source word. For each target hypothesis, a translation match score comprises a combination of the word match scores for the target hypothesis and the source words in the input series of source words.

Automatic construction of conditional exponential models from elementary features

The present invention is a system for translating text from a first source language into a second target language. The system assigns probabilities or scores to various target-language translations and then displays or makes otherwise available the highest scoring translations. The source text is first transduced into one or more intermediate structural representations. From these intermediate source structures a set of intermediate target-structure hypotheses is generated. These hypotheses are scored by two different models: a language model which assigns a probability or score to an intermediate target structure, and a translation model which assigns a probability or score to the event that an intermediate target structure is translated into an intermediate source structure. Scores from the translation model and language model are combined into a combined score for each intermediate target-structure hypothesis. Finally, a set of target-text hypotheses is produced by transducing the highest scoring target-structure hypotheses into portions of text in the target language. The system can either run in batch mode, in which case it translates source-language text into a target language without human assistance, or it can function as an aid to a human translator. When functioning as an aid to a human translator, the human may simply select from the various translation hypotheses provided by the system, or he may optionally provide hints or constraints on how to perform one or more of the stages of source transduction, hypothesis generation and target transduction.

System for parametric text to text language translation

Method and system for natural language translation

An apparatus for translating a series of source words in a first language to a series of target words in a second language. For an input series of source words, at least two target hypotheses, each including a series of target words, are generated. Each target word has a context comprising at least one other word in the target hypothesis. For each target hypothesis, a language model match score including an estimate of the probability of occurrence of the series of words in the target hypothesis. At least one alignment connecting each source word with at least one target word in the target hypothesis is identified. For each source word and each target hypothesis, a word match score including an estimate of the conditional probability of occurrence of the source word, given the target word in the target hypothesis which is connected to the source word and given the context in the target hypothesis of the target word which is connected to the source word. For each target hypothesis, a translation match score including a combination of the word match scores for the target hypothesis and the source words in the input series of source words. A target hypothesis match score including a combination of the language model match score for the target hypothesis and the translation match score for the target hypothesis. The target hypothesis having the best target hypothesis match score is output.

Language translation apparatus and method using context-based

A method and apparatus for modeling words based on match scores representing (a) the closeness of a match between probabilistic word models and the acoustic features of at least two utterances, and (b) the closeness of a match between word models and the spelling of the word. A match score is calculated for a selection set of one or more probabilistic word models. A match score is also calculated for an expansion set comprising the probabilistic word models in the selection set and one probabilistic word model from a candidate set. If the expansion set match score improves the selection set match score by a selected nonzero threshold value, the word is modelled with the word models in the expansion set. If the expansion set match score does not improve the selection set match score by the selected nonzero threshold value, the word is modelled with the words in the selection set.

Method and apparatus for generating models of spoken words based on a

A speech recognition system displays a source text of one or more words in a source language. The system has an acoustic processor for generating a sequence of coded representations of an utterance to be recognized. The utterance comprises a series of one or more words in a target language different from the source language. A set of one or more speech hypotheses, each comprising one or more words from the target language, are produced. Each speech hypothesis is modeled with an acoustic model. An acoustic match score for each speech hypothesis comprises an estimate of the closeness of a match between the acoustic model of the speech hypothesis and the sequence of coded representations of the utterance. A translation match score for each speech hypothesis comprises an estimate of the probability of occurrence of the speech hypothesis given the occurrence of the source text. A hypothesis score for each hypothesis comprises a combination of the acoustic match score and the translation match score. At least one word of one or more speech hypot

Speech recognition system for natural language translation

A language generator for a speech recognition apparatus scores a word-series hypothesis by combining individual scores for each word in the hypothesis. The hypothesis score for a single word comprises a combination of the estimated conditional probability of occurrence of a first class of words comprising the word being scored, given the occurrence of a context comprising the words in the word-series hypothesis other than the word being scored, and the estimated conditional probability of occurrence of the word being scored given the occurrence of the first class of words, and given the occurrence of the context. An apparatus and method are provided for classifying multiple series of words for the purpose of obtaining useful hypothesis scores in the language generator and speech recognition apparatus.

Speech recognition apparatus which predicts word classes from context

In a Markov model speech recognition system, an acoustic processor generates one label after another selected from an alphabet of labels. Each vocabulary word is represented as a baseform constructed of a sequence of Markov models. Each Markov model is stored in a computer memory as (a) a plurality of states; (b) a plurality of arcs, each extending from a state to a state with a respective stored probability; and (c) stored label output probabilities, each indicating the likelihood of a given label being produced at a certain arc. Word likelihood based on acoustic characteristics is determined by matching a string of labels generated by the acoustic processor against the probabilities stored for each word baseform. Improved models of words are obtained by specifying label parameters and constructing word baseforms interdependently and iteratively.

Automatic determination of labels and Markov word models in a speech

A continuous speech recognition system includes an automatic phonological rules generator which determines variations in the pronunciation of phonemes based on the context in which they occur. This phonological rules generator associates sequences of labels derived from vocalizations of a training text with respective phonemes inferred from the training text. These sequences are then annotated with their pheneme context from the training text and clustered into groups representing similar pronunciations of each phoneme. A decision tree is generated using the context information of the sequences to predict the clusters to which the sequences belong. The training data is processed by the decision tree to divide the sequences into leaf-groups representing similar pronunciations of each phoneme. The sequences in each leaf-group are clustered into sub-groups representing respectively different pronunciations of their corresponding phoneme in a give context. A Markov model is generated for each sub-group. The various Markov models of a leaf-group are combined into a single compound model by assigning common initial and final states to each model. The compound Markov models are used by a speech recognition system to analyze an unknown sequence of labels given its context.

Method and apparatus for the automatic determination of phonological

In order to determine a next event based upon available data, a binary decision tree is constructed having true or false questions at each node and a probability distribution of the unknown next event based upon available data at each leaf. Starting at the root of the tree, the construction process proceeds from node-to-node towards a leaf by answering the question at each node encountered and following either the true or false path depending upon the answer. The questions are phrased in terms of the available data and are designed to provide as much information as possible about the next unknown event. The process is particularly useful in speech recognition when the next word to be spoken is determined on the basis of the previously spoken words.

Design and construction of a binary-tree system for language modelling

In a word, or speech, recognition system for decoding a vocabulary word from outputs selected from an alphabet of outputs in response to a communicated word input wherein each word in the vocabulary is represented by a baseform of at least one probabilistic finite state model and wherein each probabilistic model has transition probability items and output probability items and wherein a value is stored for each of at least some probability items, the present invention relates to apparatus and method for determining probability values for probability items by biassing at least some of the stored values to enhance the likelihood that outputs generated in response to communication of a known word input are produced by the baseform for the known word relative to the respective likelihood of the generated outputs being produced by the baseform for at least one other word. Specifically, the current values of counts --from which probability items are derived--are adjusted by uttering a known word and determining how often probability events occur relative to (a) the model corresponding to the known uttered 'correct' word and (b) the model of at least one other 'incorrect' word. The current count values are increased based on the event occurrences relating to the correct word and are reduced based on the event occurrences relating to the incorrect word or words.

Growing community of inventors

New York, NY, United States of America

Peter F Brown