Abstract: Portions from segment boundary regions of a plurality of speech segments are extracted. Each segment boundary region is based on a corresponding initial unit boundary. Feature vectors that represent the portions in a vector space are created. For each of a plurality of potential unit boundaries within each segment boundary region, an average discontinuity based on distances between the feature vectors is determined. For each segment, the potential unit boundary associated with a minimum average discontinuity is selected as a new unit boundary.

Abstract: Portions from time-domain speech segments are extracted. Feature vectors that represent the portions in a vector space are created. The feature vectors incorporate phase information of the portions. A distance between the feature vectors in the vector space is determined. In one aspect, the feature vectors are created by constructing a matrix W from the portions and decomposing the matrix W. In one aspect, decomposing the matrix W comprises extracting global boundary-centric features from the portions. In one aspect, the portions include at least one pitch period. In another aspect, the portions include centered pitch periods.

Abstract: Exemplary embodiments of methods and apparatuses for automatic speech recognition are described. First model parameters associated with a first representation of an input signal are generated. The first representation of the input signal is a discrete parameter representation. Second model parameters associated with a second representation of the input signal are generated. The second representation of the input signal includes a continuous parameter representation of residuals of the input signal. The first representation of the input signal includes discrete parameters representing first portions of the input signal. The second representation includes discrete parameters representing second portions of the input signal that are smaller than the first portions. Third model parameters are generated to couple the first representation of the input signal with the second representation of the input signal. The first representation and the second representation of the input signal are mapped into a vector space.

Abstract: A method and apparatus for filtering messages comprising determining a first semantic anchor corresponding to a first group of messages, for example, legitimate messages and a second semantic anchor corresponding to a second group of messages, for example, unsolicited messages. Determining a vector corresponding to an incoming message; comparing the vector corresponding to the incoming message with at least one of the first semantic anchor and the second semantic anchor to obtain a first comparison value and a second comparison value; and filtering the incoming message based on the first comparison value and the second comparison value.

Abstract: A method, apparatus, and signal-bearing medium that files data in a destination based on one or more criteria. In various embodiments, the data may be email, email attachments, faxes, telephone messages, downloaded data or programs, audio, video, scanned images, photographs, blocks of text, or other data. In an embodiment, a training mode and an automatic mode are provided. During the training mode, a user is presented with data and a recommended destination, and the user provides feedback that is used to train the criteria. During an automatic mode, the data may be transferred to the destination with or without user confirmation.

Abstract: A method and apparatus for filtering messages comprising determining a first semantic anchor corresponding to a first group of messages, for example, legitimate messages and a second semantic anchor corresponding to a second group of messages, for example, unsolicited messages. Determining a vector corresponding to an incoming message; comparing the vector corresponding to the incoming message with at least one of the first semantic anchor and the second semantic anchor to obtain a first comparison value and a second comparison value; and filtering the incoming message based on the first comparison value and the second comparison value.

Abstract: A method and apparatus is provided for generating speech that sounds more natural. Determining whether information in a current sentence is new or previously given is performed based on a semantic relationship between the current sentence and a number of preceding sentences. A word prominence for the synthetic speech to a word in the current sentence is assigned in accordance with the information determination. A speech representative of the current sentence can be generated. In one embodiment, word prominence and latent semantic analysis are used to generate more natural sounding speech. A method for generating speech that sounds more natural may comprise generating synthesized speech having certain word prominence characteristics and applying a semantically-driven word prominence assignment model to specify word prominence consistent with the way humans assign word prominence.

Abstract: Portions from time-domain speech segments are extracted. Feature vectors that represent the portions in a vector space are created. The feature vectors incorporate phase information of the portions. A distance between the feature vectors in the vector space is determined. In one aspect, the feature vectors are created by constructing a matrix W from the portions and decomposing the matrix W. In one aspect, decomposing the matrix W comprises extracting global boundary-centric features from the portions. In one aspect, the portions include at least one pitch period. In another aspect, the portions include centered pitch periods.

Abstract: A method and system for dynamic language modeling of a document are described. In one embodiment, a number of local probabilities of a current document are computed and a vector representation of the current document in a latent semantic analysis (LSA) space is determined. In addition, a number of global probabilities based upon the vector representation of the current document in an LSA space is computed. Further, the local probabilities and the global probabilities are combined to produce the language modeling.

Abstract: Pronunciation for an input word is modeled by generating a set of candidate phoneme strings having pronunciations close to the input word in an orthographic space. Phoneme sub-strings in the set are selected as the pronunciation. In one aspect, a first closeness measure between phoneme strings for words chosen from a dictionary and contexts within the input word is used to determine the candidate phoneme strings. The words are chosen from the dictionary based on a second closeness measure between a representation of the input word in the orthographic space and orthographic anchors corresponding to the words in the dictionary. In another aspect, the phoneme sub-strings are selected by aligning the candidate phoneme strings on common phoneme sub-strings to produce an occurrence count, which is used to choose the phoneme sub-strings for the pronunciation.

Abstract: Portions from time-domain speech segments are extracted. Feature vectors that represent the portions in a vector space are created. The feature vectors incorporate phase information of the portions. A distance between the feature vectors in the vector space is determined. In one aspect, the feature vectors are created by constructing a matrix W from the portions and decomposing the matrix W. In one aspect, decomposing the matrix W comprises extracting global boundary-centric features from the portions. In one aspect, the portions include at least one pitch period. In another aspect, the portions include centered pitch periods.

Abstract: A method, apparatus, and signal-bearing medium that filter data based on a criteria. In an embodiment, the criteria may be related to filtering out unwanted or junk input data. In another embodiment, the criteria may be related to filtering based on desired data. In various embodiments, the data may be email, email attachments, faxes, popup windows, telephone messages, downloaded data or programs, image data, or other data. In a embodiment, a training mode and an automatic mode are provided. During the training mode, a user may be presented with data that may be junk, and feedback may be provided that is used to train a junk filter. During an automatic mode, junk data may be removed from view, transferred to a junk box, or highlighted.

Abstract: Portions from segment boundary regions of a plurality of speech segments are extracted. Each segment boundary region is based on a corresponding initial unit boundary. Feature vectors that represent the portions in a vector space are created. For each of a plurality of potential unit boundaries within each segment boundary region, an average discontinuity based on distances between the feature vectors is determined. For each segment, the potential unit boundary associated with a minimum average discontinuity is selected as a new unit boundary.

Abstract: Portions from segment boundary regions of a plurality of speech segments are extracted. Each segment boundary region is based on a corresponding initial unit boundary. Feature vectors that represent the portions in a vector space are created. For each of a plurality of potential unit boundaries within each segment boundary region, an average discontinuity based on distances between the feature vectors is determined. For each segment, the potential unit boundary associated with a minimum average discontinuity is selected as a new unit boundary.

Abstract: The present invention provides, among other things, automatic identification of near-redundant units in a large TTS voice table, identifying which units are distinctive enough to keep and which units are sufficiently redundant to discard. According to an aspect of the invention, pruning is treated as a clustering problem in a suitable feature space. All instances of a given unit (e.g. word or characters expressed as Unicode strings) are mapped onto the feature space, and cluster units in that space using a suitable similarity measure. Since all units in a given cluster are, by construction, closely related from the point of view of the measure used, they are suitably redundant and can be replaced by a single instance. The disclosed method can detect near-redundancy in TTS units in a completely unsupervised manner, based on an original feature extraction and clustering strategy.

Abstract: An orthographic anchor for each word in a dictionary is created in an orthographic space by mapping the words and a set of letter patterns characteristic of the words into the orthographic space. In one aspect the orthographic anchors are row or column vectors resulting from a decomposition of a matrix of feature vectors created by the mapping. In another aspect, a pronunciation for an input word is modeled based on a set of candidate phoneme strings that have pronunciations close to the input word in the orthographic space.

Abstract: A method and apparatus is provided for generating speech that sounds more natural. In one embodiment, word prominence and latent semantic analysis are used to generate more natural sounding speech. A method for generating speech that sounds more natural may comprise generating synthesized speech having certain word prominence characteristics and applying a semantically-driven word prominence assignment model to specify word prominence consistent with the way humans assign word prominence. A speech representative of a current sentence is generated. The determination is made whether information in the current sentence is new or previously given in accordance with a semantic relationship between the current sentence and a number of preceding sentences. A word prominence is assigned to a word in the current sentence in accordance with the information determination.

Abstract: An extended finite state grammar structure is generated from a finite state grammar. The extended finite state grammar structure includes word subgraphs representing a set of pre-defined word strings for words in the finite state grammar, and a set of all possible word strings for the words. The extended finite state grammar structure can be used to transform audio input into one or more of the word strings.

Abstract: Methods, apparatus, systems, and computer program products are provided for synthesizing speech. One method includes matching a first level of units of a received input string to audio segments from a plurality of audio segments including using properties of or between first level units to locate matching audio segments from a plurality of selections, parsing unmatched first level units into second level units, matching the second level units to audio segments using properties of or between the units to locate matching audio segments from a plurality of selections and synthesizing the input string, including combining the audio segments associated with the first and second units.

Abstract: A method and system for dynamic language modeling of a document are described. In one embodiment, a number of local probabilities of a current document are computed and a vector representation of the current document in a latent semantic analysis (LSA) space is determined. In addition, a number of global probabilities based upon the vector representation of the current document in an LSA space is computed. Further, the local probabilities and the global probabilities are combined to produce the language modeling.