Abstract: A method and apparatus for adapting reference templates is provided. The method includes adapting one or more reference templates using a stored test utterance by replacing data within at the reference templates with a weighted interpolation of that data and corresponding data within the test utterance.

Abstract: An audio signal (172) representative of an acoustic signal is provided to an auditory model (105). The auditory model (105) produces a high-dimensional feature set based on physiological responses, as simulated by the auditory model (105), to the acoustic signal. A multidimensional analyzer (106) orthogonalizes and truncates the feature set based on contributions by components of the orthogonal set to a cortical representation of the acoustic signal. The truncated feature set is then provided to classifier (108), where a predetermined sound is discriminated from the acoustic signal.

Abstract: The delay time and listening quality of a system under test are measured from a signal received therefrom, then the measured delay time and listening quality are transformed to a delay-related degradation and a listening quality degradation on the same quality measure, then the quantity of interaction between the delay-related degradation and the listening quality degradation is calculated, and the delay-related degradation, the listening quality degradation and the quantity of interaction are added together to obtain an overall degradation. The overall degradation is transformed to a subjective evaluation value to estimate the overall speech quality.

Abstract: A method and system provide for selecting a web site home page in a desired language. The method comprises identifying within a web address request a directional information item by extracting from the request a site language cookie stored in a user's access device, providing a web page associated with the web address to a second web site corresponding to the directional information item wherein the second web site includes a language translator, translating the web page textual information in accordance with the language translator and returning the translated web page to the user. The method further provides for accessing a control table to determine a status of the web page and obtaining a version of the web page stored locally on the second site or otherwise obtaining a current version of the web page; and translating the obtained web page.

Abstract: The invention concerns audio coding methods and particularly relates to an efficient means by which selected frequency bands of information from an original audio signal which are audible but which are perceptually less relevant need not be encoded, but may be replaced by a noise filling parameter. Those signal bands having content which is perceptually more relevant are, in contrast fully encoded. Encoding bits may be saved in this manner, without leaving voids in the frequency spectrum of the received signal. In this way, this method avoids the annoying bandwidth switching artefacts that can occur when full bandwidth audio is encoded with a bit budget which is too low to represent the signal within each frequency band. Thus, this method allows an increase in the encoded audio bandwidth without introducing annoying bandwidth switching artefacts.

Abstract: A speech recognition (SR) dictionary switching unit selects a SR dictionary including commands that are for controlling target devices and selectable in current operating states detected by a device-state detecting unit. When a SR engine recognizes an uttered command of a user, it collates the speech with the commands included in the selected SR dictionary. If the command is not selectable in the current operating states of the target devices and is one of a reciprocal pair of commands working oppositely to each other, the other of the reciprocal pair is substituted for the command. This results in decreasing unreasonable and unpleasant feeling of the user in voice manipulation.

Abstract: Embodiments of the present invention relate to a method and system for augmenting a training database of an automated language-understanding system. In one embodiment, a training example in a first language is received from the training database. The first language-training example translated to a second language output. The second language output is translated to a first variant of the first language-training example. An action pair including the first variant of the first language-training example and an action command associated with the first language-training example is stored in an augmented training database.

Abstract: A method for translating medical test results into plain language is provided. The method includes receiving a medical test result for a type of medical test and, after making a threshold determination whether the medical test result will initially be automatically interpreted by the computer system independent of clinician input, identifying a template or set of templates associated with the type of medical test. The method also includes selecting the template matching the medical test result and outputting a plain language explanation based on the selected template.

Abstract: A method and system for analyzing natural language text to identify a set of categories of data in a knowledge base associated with the natural language text and selecting a category from the set of categories is provided. The method and system further provides automatically providing a single category if the set of categories only contains the single category. The method and system also provides that each category in the knowledge base is associated with a stored text representation of a sample natural language text and a score for the text representation-category combination, wherein the score is representative of a relative relevance of the text representation-category combination. The method and system further provide for modifying the score associated with the text representation-category combination of a selected category using feedback provided upon the selection.

Abstract: A neural network is used to obtain more robust performance in determining prosodic markers on the basis of linguistic categories.

Abstract: A system and method for inferring informational goals and preferred level of details in answers in response to questions posed to computer-based information retrieval or question-answering systems is provided. The system includes a query subsystem that can receive an input query and extrinsic data associated with the query and which can output an answer to the query. The query subsystem accesses an inference model to retrieve conditional probabilities that certain informational goals are present. One application of the system includes determining a user's likely informational goals and then accessing a knowledge data store to retrieve responsive information. Determining a user's likely informational goals can include inferring a desired level of detail of answers to the query based on the application being employed by the user at the time the query is submitted.

Abstract: In accordance with the present invention, a method for increasing both dialect precision and usability in speech recognition and text-to-speech systems is described. The invention generates non-linear (i.e. encoded)baseform representations for words and phrases from a pronunciation lexicon. The baseform representations are encoded to incorporate both pronunciation variations and dialectal variations. The encoded baseform representations may be later expanded (i.e. decoded) into one or more linear dialect specific baseform representations, utilizing a set of dialect specific phonological rules. The method comprises the steps of: constructing an encoded pronunciation lexicon having a plurality of encoded and unencoded baseforms; inputting one or more user specified dialects; selecting dialect specific phonological rules from a rule set database; and decoding the encoded pronunciation lexicon using the dialect specific phonological rules to yield a dialect specific decoded pronunciation lexicon.

Abstract: Broadcast audio is automatically monitored for information of interest. A computer processor has a memory associated therewith for storing a vocabulary of keywords of interest. An audio receiver receives an audio broadcast. A speech recognition system associated with the audio receiver and computer processor detects when one of the keywords appears in a received audio segment. A report generator associated with the computer processor and responsive to the detection of a keyword generates a report detailing the detected keyword and its context. The system is particularly well suited to the verification of commercial messages and to assist in public relations efforts.

Abstract: A method for computing a distance between collections of distributions or finite mixture models of features. Data is processed so as to define at least first and second collections of distributions of features. For each distribution of the first collection, the distance to each distribution of the second collection is measured to determine which distribution of the second collection is the closest (most similar). The same procedure is performed for the distributions of the second collection. Based on the closest distance measures, a final distance is computed representing the distance between the first and second collections. This final distance may be a weighted sum of the closest distances. The distance measure may be used in a number of applications such as [speaker classification,] speaker recognition and audio segmentation.

Abstract: A CMOS implementation for a DC holding circuit in direct access arrangement (DAA) circuitry is disclosed that provides desirable inductive behavior while minimizing power dissipation required by the CMOS integrated circuit, particularly at high loop currents. The DC holding circuitry disclosed may include MOS transistors located on a CMOS integrated circuit and an off-chip power dissipating resistor that acts to dissipate power external to the CMOS integrated circuit. The CMOS implementation disclosed also allows a path for drawing DC current to power other CMOS circuits (e.g., ADCs and DACs) in the CMOS integrated circuit.

Abstract: A computer system with speech recognition systems and handwriting recognition systems are disclosed that work closely together to improve the total recognition accuracy of each system alone. Handwriting recognition system may include a pen/stylus input device and associated program functions. These systems or programs may be combined with computer and telephone functions to provide integrated applications having voice output programs, Internet access, e-mail/v-mail, and personal information manager functions. The computer system can recognize speaker-dependent and speaker-independent speech, converting this information to computer recognizable text, which may be displayed onto a display device in realtime or delayed. Speech recognition errors may be corrected via a pen input device, and the pen information may be recognized, converted to text and graphics. This data may then be displayed immediately or displayed later at a user specified time.

Abstract: Language model results are combined according to a combination expression to produce combined language model results for a set of candidates. A candidate is selected and the combination expression is adjusted using language model results associated with the selected candidate.

Abstract: A speech recognition synthesis based encoding/decoding method recognizes phonetic segments, syllables, words or the like as character information from an input speech signal and detects pitch periods, phoneme or syllable durations or the like, as information for prosody generation, from the input speech signal, transfers or stores the character information and information for prosody generation as code data, decodes the transferred or stored code data to acquire the character information and information for prosody generation, and synthesizes the acquired character information and information for prosody generation to obtain a speech signal.

Abstract: Information (e.g. traffic information) is retrieved from a server. The content of the information is reviewed in order to determine which of a plurality of prestored speech files should be used to report the information. The selected speech files are concatenated to form an audio presentation. The speech files may include phrases identifying incident types, locations, severity and/or timing information, as well as other filler words that improve the audio presentation. In some embodiments, the audio presentation is accompanied by video and/or graphics that can be referenced by the audio presentation.

Abstract: The communication system disclosed herein allows for the hookswitch devices to draw loop current from the phone line in both on-hook and off-hook conditions. Thus, even though an on-hook condition occurs, current may be obtained through the hookswitch devices. This feature allows circuitry which operates during on-hook conditions to still receive power from the phone line. Moreover because the hookswitch devices are utilized for drawing power in both on-hook and off-hook conditions, the use of additional switches dedicated to drawing the power during on-hook conditions is not required. For example, loop current may be drawn from the phone line through the hookswitch devices to power circuits used to perform the on-hook caller ID function. The powered circuits may include for example analog to digital converters. The technique disclosed herein for drawing current through the hookswitch devices is particularly useful for communications systems which utilize a capacitive isolation barrier system.