Abstract: An audio decoding method and apparatus for reconstructing high frequency components with less computation are provided. The audio decoding apparatus includes a decoder, a channel similarity determination unit, a high frequency component generation unit, and an audio synthesizing unit. The audio decoding method generates high frequency components of frames while skipping every other frame for each channel signal; when right and left channel signals are similar to each other, generates high frequency components of the skipped frame for any one channel signal by using the generated high frequency components of the corresponding frame for the other channel signal; and when the right and left channel signals are not similar to each other, generates high frequency components of the skipped frames for each channel signal by using previous frames for the relevant channel signal.

Abstract: In a semi-automatic construction method for knowledge base of an encyclopedia question answering system, concept-oriented systematic templates are designed and important fact information related to entries is automatically extracted from summary information and body of the encyclopedia to semi-automatically construct the knowledge base of the encyclopedia question answering system. The method includes the steps of: (a) designing structure of the knowledge base with a plurality of templates for each entry and a plurality attributes related to each of the templates; (b) extracting structured information including the entry, an attribute name and attribute values from summary information of the encyclopedia; (c) extracting unstructured information including an attribute name and attribute values of the entry from a body of the encyclopedia; and (d) storing the structured information and the unstructured information in corresponding template and attribute of the knowledge base according to the entry.

Abstract: A method for identifying common multiphone units to add to a unit inventory for a text-to-speech generator is disclosed. The common multiphone units are units that are larger than a phone, but smaller than a syllable. The method slices each syllable into a plurality of slices. These slices are then sorted and the frequency of each slice is determined. Those slices whose frequencies exceed a threshold are added to the unit inventory. The remaining slices are decomposed according to a predetermined set of rules to determine if they contain slices that should be added to the unit inventory.

Abstract: A data input device for inputting numeric data by voice includes a range prediction part, a history holding part, a speech recognition part, a recognition result holding part, a comparison part, a presentation part, and a result storing part. The range prediction part estimates a range of a value expected to be input on the basis of meter-reading history data held in the history holding part. The speech recognition part recognizes speech representing a meter reading and stores the recognition result in the recognition result holding part. The comparison part determines whether or not the meter reading for this month represented by the data stored in the recognition result holding part is within the prediction range. If the meter reading for this month is within the prediction range, the presentation part presents the recognition result to a user, and the speech recognition result is stored in the result storing part.

Abstract: A directed dialogue portal system management scheme is disclosed, providing a manager application layer over portal system managed applications. The manager application layer may include a global grammar (e.g., the grammar of each managed application) and each managed application may inherit behaviors of the manager application. Responses to prompts received by a first managed application that are unrecognized or are directed to a second managed application are sent to the manager application. The manager application sends the response to the second application. The scheme is transparent to the user and, from the user's perspective, a user is able to directly access any application or services in a directed dialogue portal system from other locations in the system.

Abstract: A method and system for automatically generating a scoring model for scoring a speech sample are disclosed. One or more training speech samples are received in response to a prompt. One or more speech features are determined for each of the training speech samples. A scoring model is then generated based on the speech features. At least one of the training speech samples may be a high entropy speech sample. An evaluation speech sample is received and a score is assigned to the evaluation speech sample using the scoring model. The evaluation speech sample may be a high entropy speech sample.

Abstract: A system and method for effectively implementing an optimized language model for speech recognition includes initial language models each created by combining source models according to selectable interpolation coefficients that define proportional relationships for combining the source models. A rescoring module iteratively utilizes the initial language models to process input development data for calculating word-error rates that each correspond to a different one of the initial language models. An optimized language model is then selected from the initial language models by identifying an optimal word-error rate from among the foregoing word-error rates. The speech recognizer may then utilize the optimized language model for effectively performing various speech recognition procedures.

Abstract: A method for unifying speech user interface and graphic user interface commands includes the steps of receiving (52) user entered text via a GUI, processing (54) the user-entered text, monitoring (60) the user-entered text, adding input context (62) to the user-entered text, and, updating (74, 76, and 78) a speech recognizer with the user-entered text and semantic information. Updating the speech recognizer can include the step of accepting new text information and input context to update a speech grammar (74) and recognition vocabulary of the speech recognizer. The method can include the step of updating the GUI (72) by updating GUI directives (68) and elements (70) to maintain the GUI unified with the speech grammar. The method can further include the step of forming a window (402) enabling the display of a speech interface command as a user constructs the speech interface command using the GUI (400).

Abstract: A system and method for identifying language attributes through probabilistic analysis is described. A set of language classes and a plurality of training documents are defined, Each language class identifies a language and a character set encoding. Occurrences of one or more document properties within each training document are evaluated. For each language class, a probability for the document properties set conditioned on the occurrence of the language class is calculated. Byte occurrences within each training document are evaluated. For each language class, a probability for the byte occurrences conditioned on the occurrence of the language class is calculated.

Abstract: Semantic understanding of hypotheses returned by a speech engine could improve the quality of recognition and in cases of misrecognition speed the identification of errors and potential substitutions. Unfortunately, semantic recognition using natural language parsers is hard since semantic and syntactic rules for processing language are complex and computationally expensive. Additionally, semantic recognition should be performed in a knowledge domain—a domain of interest such as radiology, pathology, or tort law. This adds additional complexity to building semantic rules. The method described achieves semantic understanding by coupling a speech recognition engine to a semantic recognizer, which draws from a database of domain sentences derived from a document corpus, and a knowledge base created for these domain sentences. The method is able to identify in near real-time the best sentence hypotheses from the speech recognizer and its associated meanings, i.e.

Abstract: To provide an acoustic model which can absorb the fluctuation of a phonemic environment in an interval longer than a syllable, with the number of parameters of the acoustic model suppressed to be small, a phoneme-connected syllable HMM/syllable-connected HMM set is generated in such a way that a phoneme-connected syllable HMM set corresponding to individual syllables is generated by combining phoneme HMMs. A preliminary experiment is conducted using the phoneme-connected syllable HMM set and training speech data. Any misrecognized syllable and the preceding syllable of the misrecognized syllable are checked using results of a preliminary experiment syllable label data. The combination between a correct answer syllable for the misrecognized syllable and the preceding syllable of the misrecognized syllable is extracted as a syllable connection. A syllable-connected HMM corresponding to this syllable connection is added into the phoneme-connected syllable HMM set.

Abstract: A method for recovering interrupted voice input, particularly voice input into communication, audio and/or navigation systems installed in a vehicle detects, during an ongoing voice input process, an event that requires interruption of the voice input process, stores an incomplete sequence of the voice input, and interrupts the voice input process in response to detection of the event. The method can further detect that the event has finished and can recover the voice input process based on the stored sequence. A corresponding device is also described.

Abstract: A method for processing language input can include the step of determining at least two possible meanings for a language input. For each possible meaning, a probability that the possible meaning is a correct interpretation of the language input can be determined. At least one relative data computation can be computed based at least in part upon the probabilities. At least one irregularity within the language input can be detected based upon the relative delta computation. The irregularity can include mumble, ambiguous input, and/or compound input. At least one programmatic action can be performed responsive to the detection of the irregularity.

Abstract: A method and system for dynamically assigning weights to the subset of commands in a natural language dialog system based on prior context of the user's interaction with the system. The search space of the translation process may be reduced when some context information is available. A user presents input to the natural language understanding system. The system translates the user input into a formal command and calculates a weight value for a next set of formal commands based on the formal command. The command weights may then be dynamically boosted for the next set of formal commands before executing the formal command. The exemplary aspects of the present invention reduce the time needed to complete a task since the search space of the translation process may be reduced if context information is available and improve the accuracy of the process by using knowledge that users regularly use repeating patterns for repeating tasks.

Abstract: A speech processing apparatus able to enhance formants more naturally, wherein a speech analyzing unit analyzes an input speech signal to find LPCs and converts the LPCs to LSPs, a speech decoding unit calculates a distance between adjacent orders of the LSPs by an LSP analytical processing unit and calculates LSP adjusting amounts of larger values for LSPs of adjacent orders closer in distance by an LSP adjusting amount calculating unit, an LSP adjusting unit adjusts the LSPs based on the LSP adjusting amounts such that the LSPs of adjacent orders closer in distance become closer, an LSP-LPC converting unit converts the adjusted LSPs to LPCs, and an LPC combining unit uses the LPCs and sound source parameters to obtain formant-enhanced speech.

Abstract: A system and method for a speech recognition technology that allows language models for a particular language to be customized through the addition of alternate pronunciations that are specific to the accent of the dictator, for a subset of the words in the language model. The system includes the steps of identifying the pronunciation differences that are best handled by modifying the pronunciations of the language model, identifying target words in the language model for pronunciation modification, and creating a accented speech file used to modify the language model.

Abstract: A method of smoothing fundamental frequency discontinuities at boundaries of concatenated speech segments includes determining, for each speech segment, a beginning fundamental frequency value and an ending fundamental frequency value. The method further includes adjusting the fundamental frequency contour of each of the speech segments according to a linear function calculated for each particular speech segment, and dependent on the beginning and ending fundamental frequency values of the corresponding speech segment. The method calculates the linear function for each speech segment according to a coupled spring model with three springs for each segment. A first spring constant, associated with the first spring and the second spring, is proportional to a duration of voicing in the associated speech segment. A second spring constant, associated with the third spring, models a non-linear restoring force that resists a change in slope of the segment fundamental frequency contour.

Abstract: In a method of assisting a subject to generate speech, at least one first neural impulse is sensed from a first preselected location in the subject's brain. A first preselected sound is associated with the first neural impulse. The first preselected sound is generated in an audible format. In an apparatus for assisting the subject to generate speech, at least one sensor senses a neural impulse in the subject's brain and generates a signal representative thereof. An electronic speech generator generates a phoneme in response to the generation of the signal. An audio system generates audible sounds corresponding to the phoneme based upon the signal received from the speech generator.