Abstract: Techniques for combining the results of multiple recognizers in a distributed speech recognition architecture. Speech data input to a client device is encoded and processed both locally and remotely by different recognizers configured to be proficient at different speech recognition tasks. The client/server architecture is configurable to enable network providers to specify a policy directed to a trade-off between reducing recognition latency perceived by a user and usage of network resources. The results of the local and remote speech recognition engines are combined based, at least in part, on logic stored by one or more components of the client/server architecture.

Abstract: An error concealment method and apparatus for an audio signal and a decoding method and apparatus for an audio signal using the error concealment method and apparatus. The error concealment method includes selecting one of an error concealment in a frequency domain and an error concealment in a time domain as an error concealment scheme for a current frame based on a predetermined criteria when an error occurs in the current frame, selecting one of a repetition scheme and an interpolation scheme in the frequency domain as the error concealment scheme for the current frame based on a predetermined criteria when the error concealment in the frequency domain is selected, and concealing the error of the current frame using the selected scheme.

Abstract: IP telephony communications are conducted by sending both data produced by a CODEC that represents received spoken audio input, and a textual representation of the spoken audio input. A receiving device utilizes the textual representation of the spoken audio input to help recreate the spoken audio input when a portion of the CODEC data is missing. The textual representation can be generated by a speech-to-text function. Alternatively, the textual representation can be a notation of extracted phonemes.

Abstract: A caption translation system is described herein that provides a way to reach a greater world-wide audience when displaying video content by providing dynamically translated captions based on the language the user has selected for their browser. The system provides machine-translated captions to accompany the video content by determining the language the user has selected for their browser or a manual language selection of the user. The system uses the language value to invoke an automated translation application-programming interface that returns translated caption text in the selected language. The system can use one or more well-known caption formats to store the translated captions, so that video playing applications that know how to consume captions can automatically display the translated captions. The video playing application plays back the video file and displays captions in the user's language.

Abstract: A method for extending a bandwidth of a speech signal received, according to an embodiment of the present invention, includes: transforming the received speech signal into a frequency domain by decoding the received speech signal; normalizing the transformed speech signal; differentiating a voiced sound period or unvoiced sound period from the received speech signal; extracting, from the normalized speech signal, a first period including a harmonic component of the voiced sound period on the basis of the voiced sound period; extracting, from the normalized speech signal, a second period on the basis of correlation between the unvoiced sound period and the normalized speech signal; generating a high-band speech signal on the basis of the first period and the second period; and synthesizing the generated high-band speech signal and the transformed speech signal to output a wideband speech signal.

Abstract: A system and method are provided for accelerating machine reading of text. In one embodiment, the system comprises at least one processor device. The processor device is configured to receive at least one image of text to be audibly read. The text includes a first portion and a second portion. The processor device is further configured to initiate optical character recognition (OCR) to recognize the first portion. The processor device is further configured to initiate an audible presentation of the first portion prior to initiating OCR of the second portion, and simultaneously perform OCR to recognize the second portion of the text to be audibly read during presentation of at least part of the first portion. The processor device is further configured to automatically cause the second portion of the text to be audibly presented immediately upon completion of the presentation of the first portion.

Abstract: A textual message processing system and method are described for use in a mobile environment. A user messaging application processes at least one user textual message during a user messaging session. A semantic annotation module identifies one or more semantically salient terms in the user textual message, and annotates the user textual message with annotation terms having a low semantic distance to the semantically salient terms. A user message history stores the annotated textual messages. The semantic annotation module may further annotate the user textual message with situational meta-data characterizing the user textual message. There may be a message search module for using one or more keywords to search the user message history including the annotation terms, and identifying as a search match any annotated textual messages within a semantic distance threshold of the one or more keywords.

Abstract: Techniques for combining the results of multiple recognizers in a distributed speech recognition architecture. Speech data input to a client device is encoded and processed both locally and remotely by different recognizers configured to be proficient at different speech recognition tasks. The client/server architecture is configurable to enable network providers to specify a policy directed to a trade-off between reducing recognition latency perceived by a user and usage of network resources. The results of the local and remote speech recognition engines are combined based, at least in part, on logic stored by one or more components of the client/server architecture.

Abstract: In an automated Question Answer (QA) system architecture for automatic open-domain Question Answering, a system, method and computer program product for predicting the Lexical Answer Type (LAT) of a question. The approach is completely unsupervised and is based on a large-scale lexical knowledge base automatically extracted from a Web corpus. This approach for predicting the LAT can be implemented as a specific subtask of a QA process, and/or used for general purpose knowledge acquisition tasks such as frame induction from text.

Abstract: An apparatus comprising an ingress port configured to receive a signal comprising a plurality of encoded audio signals corresponding to a plurality of sources; and a processor coupled to the ingress port and configured to calculate a parameter for each of the plurality of encoded audio signals, wherein each parameter is calculated without decoding any of the encoded audio signals, select some, but not all, of the plurality of encoded audio signals according to the parameter for each of the encoded audio signals, decode the selected signals to generate a plurality of decoded audio signals, and combine the plurality of decoded audio signals into a first audio signal.

Abstract: Enhanced speech is produced from a mixed signal including noise and the speech. The noise in the mixed signal is estimated using a vector-Taylor series. The estimated noise is in terms of a minimum mean-squared error. Then, the noise is subtracted from the mixed signal to obtain the enhanced speech.

Abstract: Techniques are described herein that suppress noise using multiple sensors (e.g., microphones) of a communication device. Noise modeling (e.g., estimation of noise basis vectors and noise weighting vectors) is performed with respect to a noise signal during operation of a communication device to provide a noise model. The noise model includes noise basis vectors and noise coefficients that represent noise provided by audio sources other than a user of the communication device. Speech modeling (e.g., estimation of speech basis vectors and speech weighting) is performed to provide a speech model. The speech model includes speech basis vectors and speech coefficients that represent speech of the user. A noisy speech signal is processed using the noise basis vectors, the noise coefficients, the speech basis vectors, and the speech coefficients to provide a clean speech signal.

Abstract: A method for storing audio data is disclosed, including: recording basic information of a versatile audio data storage file into the versatile audio data storage file; storing Versatile Audio Codec (VAC) frame data into the versatile audio data storage file sequentially; recording payload information of the versatile audio data storage file into the versatile audio data storage file; and recording index information of VAC frames stored in the versatile audio data storage file into the versatile audio data storage file. A device for storing the audio data is also disclosed, including: a basic information record module, a VAC frame data storage module, a payload information record module and an index information record module. The file generated with this method is simple and is easy to read and access, which can be applied to various applications of the versatile audio frequently.

Abstract: A method of providing speech transcription performance indication includes receiving, at a user device data representing text transcribed from an audio stream by an ASR system, and data representing a metric associated with the audio stream; displaying, via the user device, said text; and via the user device, providing, in user-perceptible form, an indicator of said metric. Another method includes displaying, by a user device, text transcribed from an audio stream by an ASR system; and via the user device, providing, in user-perceptible form, an indicator of a level of background noise of the audio stream. Another method includes receiving data representing an audio stream; converting said data representing an audio stream to text via an ASR system; determining a metric associated with the audio stream; transmitting data representing said text to a user device; and transmitting data representing said metric to the user device.

Abstract: Disclosed communication network enabled system and method for connecting pluralities of users for translating information sent over a communication network comprising a mobile application installed in a portable mobile communication device for receiving the information from the users. A first time user can specify input language and the desired translated output language into the mobile application. The translation request notification information is routed and sent to top tier users selected from a ranked list by a server as a push notification. The first one to respond is connected to the user. The translator can set the frequency of translation requests and can charge for each translation. After the translation is completed the user can rate the translator which will help the translator to get new requests. Too many bad reports about the translation of a user will get that user blocked.

Abstract: A system and method for automatically generating sentences in a language is disclosed. The system comprising a grammar processor for converting an input grammar into a hierarchical representation, and a grammar explorer module for traversing the grammar hierarchy based on an explore specification, which defines what nodes of the hierarchy should be explored. The explorer module takes the exploration specification as input and traverses the hierarchy according to the exploration types specified in the exploration specification. The system and method can be used to automatically generate assembly instructions for a microprocessor given its assembly language grammar, to generate sentences of a natural language like English from its grammar and to generate programs in a high-level programming language like C.

Abstract: A method for implementing at least one rule for an application is described. The method includes receiving an input rule. Based on the input rule, a program executable code is generated. The generated program executable code can then be associated with the application.

Abstract: A system for the control of an implant (32) in a body (11), comprising first (10, 20) and second parts (12) which communicate with each other. The first part (10, 20) is adapted for implantation and for control of and communication with the medical implant (32), and the second part (12) is adapted to be worn on the outside of the body (11) in contact with the body and to receive control commands from a user and to transmit them to the first part (10, 20). The body (11) is used as a conductor for communication between the first (10, 20) and the second (12) parts. The second part (12) is adapted to receive and recognize voice control commands from a user and to transform them into signals which are transmitted to the first part (10, 20) via the body (11).

Abstract: Embodiments of the present invention provide a method, device, and system for classifying voice conference minutes. The method is: performing voice source locating according to audio data of the conference site so as to acquire a location of a voice source corresponding to the audio data, writing the location of the voice source into additional field information of the audio data, writing a voice activation flag into the additional field information, packaging the audio data as an audio code stream, and sending the audio code stream and the additional field information of the audio code stream to a recording server, so that the recording server classifies the audio data according to the additional field information and writes a participant identity that corresponds to the location of the voice source corresponding to the audio data into the additional field information of the audio code stream.

Abstract: Disclosed are methods and apparatus for processing linguistic expressions (e.g., opinionated text documents). The linguistic expressions are processed by, firstly, detecting topics of interest discussed in the linguistic expressions. The sentiment, or sentiments, of an originator with respect to each of the topics detected in the linguistic expressions is then assessed. The originators are then grouped (or clustered) into one or more groups based on the similarities between the originators' respective sets of detected topics and corresponding sentiments. Semantic information is then associated with a given group. Finally, for a given member of a given group, a profile is created or updated. This profile comprises attributes that may be based on a degree of membership of the given member to the given group and the semantic information associated with the given group.