Abstract: A method and apparatus is provided for two-tier noise rejection in speech recognition. The method and apparatus convert an analog speech signal into a digital signal and extract features from the digital signal. A hypothesis speech word and a hypothesis noise word are identified from respective extracted features. The features associated with the hypothesis speech word are examined in a second tier of noise rejection to determine if the features are more likely to represent noise than speech. The hypothesis speech word is replaced by a noise marker if the features are more likely to represent noise than speech.

Abstract: A computer implemented system and method of proofreading text in a computer system includes receiving text from a user into a text editing module. At least a portion of the text is converted to an audio signal. The audio signal is played through a speaker to the user to provide feedback.

Abstract: A speech recognition method provides improved modeling in recognition accuracy using hidden Markov models. During training, the method creates a senone tree for each state of each phoneme encountered in a data set of training words. All output distributions received for a selected state of a selected phoneme in the set of training words are clustered together in a root node of a senone tree. Each node of the tree beginning with the root node is divided into two nodes by asking linguistic questions regarding the phonemes immediately to the left and right of a central phoneme of a triphone. At a predetermined point, the tree creation stops, resulting in leaves representing clustered output distributions known as senones. The senone trees allow all possible triphones to be mapped into a sequence of senones simply by traversing the senone trees associated with the central phoneme of the triphone.

Abstract: A computer-implemented method of recognizing an input speech utterance compares the input speech utterance to a plurality of hidden Markov models to obtain a constrained acoustic score that reflects the probability that the hidden Markov model matches the input speech utterance. The method computes a confidence measure for each hidden Markov model that reflects the probability of the constrained acoustic score being correct. The computed confidence measure is then used to adjust the constrained acoustic score. Preferably, the confidence measure is computed based on a difference between the constrained acoustic score and an unconstrained acoustic score that is computed independently of any language context. In addition, a new confidence measure preferably is computed for each input speech frame from the input speech utterance so that the constrained acoustic score is adjusted for each input speech frame.

Abstract: A data compression system greatly compresses the stored data used by a speech recognition system employing hidden Markov models (HMM). The speech recognition system vector quantizes the acoustic space spoken by humans by dividing it into a predetermined number of acoustic features that are stored as codewords in a vector quantization (output probability) table or codebook. For each spoken word, the speech recognition system calculates an output probability value for each codeword, the output probability value representing an estimated probability that the word will be spoken using the acoustic feature associated with the codeword. The probability values are stored in an output probability table indexed by each codeword and by each word in a vocabulary. The output probability table is arranged to allow compression of the probability of values associated with each codeword based on other probability values associated with the same codeword, thereby compressing the stored output probability.

Abstract: A method and system for improving speech recognition through front-end normalization of feature vectors are provided. Speech to be recognized is spoken into a microphone, amplified by an amplifier, and converted from an analog signal to a digital signal by an analog-to-digital ("A/D") converter. The digital signal from the A/D converter is input to a feature extractor that breaks down the signal into frames of speech and then extracts a feature vector from each of the frames. The feature vector is input to an input normalizer that normalizes the vector. The input normalizer normalizes the feature vector by computing a correction vector and subtracting the correction vector from the feature vector. The correction vector is computed based on the probability of the current frame of speech being noise and based on the average noise and speech feature vectors for a current utterance and a database of utterances.