Abstract: A voice response system attempts to respond to spoken user input and to provide computer-generated responses. If the system decides it cannot provide valid responses, the current state of user session is determined and forwarded to a human operator for further action. The system maintains a recorded history of the session in the form of a dialog history log. The dialog history and information as to the reliability of past speech recognition efforts is employed in making the current state determination. The system includes formatting rules for controlling the display of information presented to the human operator.

Abstract: A Hybrid Distributed Network Language Translation (HDNLT) system having a distributed network of human and machine translators that communicate electronically and provide for the translation of material in source language. Individual translators receive a reputation that reflects their translation competency, reliability and accuracy. An individual translator's reputation is adjusted dynamically with feedback from other translators and/or comparison of their translation results to translations from those with known high reputation and to the final translation results. Additionally, translations are produced statistically, first by breaking input source text into fragments, sending each fragment redundantly to a number of translators with varying levels of reputation.

Abstract: The present invention proposes a new method for improving the performance of a real-valued filterbank based spectral envelope adjuster. By adaptively locking the gain values for adjacent channels dependent on the sign of the channels, as defined in the application, reduced aliasing is achieved. Furthermore, the grouping of the channels during gain-calculation, gives an improved energy estimate of the real valued subband signals in the filterbank.

Abstract: A transform coder leading to reduction of degradation of perceptual sound quality even if an adequate number of bits is not assigned. Candidates of a correction scale factor stored in a correction scale factor codebook are outputted one by one, and an error signal is generated by subjecting the candidate and scale factors outputted from scale factor computing sections to a predetermined operation. A judging section determines a weight vector given to a weighted error computing section depending on the sign of the error signal. The weighted error computing section computes the square of the error signal, multiplies the square of the error signal by the weight vector given from the judging section and computes a weighted squared error E. A search section determines the candidates of the correction scale factor which minimizes the weighted squared error E by a closed loop processing.

Abstract: Methods, encoders, and digital systems are provided for predictive encoding of speech parameters in which an input frame is encoded by quantizing a parameter vector of the input frame with a strongly-predictive codebook and a weakly-predictive codebook to obtain a strongly-predictive distortion and a weakly-predictive distortion, adjusting a correlation indicator based on a relative correlation of the input frame to a previous frame, wherein the correlation indicator is indicative of the strength of the correlation of previously encoded frames, and encoding the input frame with the weakly-predictive codebook unless the correlation indicator has reached a correlation threshold.

Abstract: The present invention proposes a new method for improving the performance of a real-valued filterbank based spectral envelope adjuster. By adaptively locking the gain values for adjacent channels dependent on the sign of the channels, as defined in the application, reduced aliasing is achieved. Furthermore, the grouping of the channels during gain-calculation, gives an improved energy estimate of the real valued subband signals in the filterbank.

Abstract: A voice response system attempts to respond to spoken user input and to provide computer-generated responses. If the system decides it cannot provide valid responses, the current state of user session is determined and forwarded to a human operator for further action. The system maintains a recorded history of the session in the form of a dialog history log. The dialog history and information as to the reliability of past speech recognition efforts is employed in making the current state determination. The system includes formatting rules for controlling the display of information presented to the human operator.

Abstract: There is provided a method of post-processing a speech signal. The method comprises applying a time-domain post-processing to the speech signal, using LPC coefficients, for a low-band frequency range and applying a frequency-domain post-processing to the speech signal, using MDCT coefficients, for the high-band frequency range. Applying the frequency-domain post-processing includes decoding an encoded speech signal to obtain MDCT coefficients representative of the speech signal divided into a plurality of sub-bands, generating an envelope for each sub-band of the plurality of sub-bands as an average magnitude of the MDCT coefficients of the sub-band, generating an envelope modification factor for each sub-band of the plurality of sub-band using the MDCT coefficients of the sub-band, modifying the envelope by the envelope modification factor for each sub-band of the plurality of sub-bands to provide a modified envelope, and generating the post-processed speech signal using the modified envelope.

Abstract: A method and apparatus for transforming an audio signal, a method and apparatus for adaptively encoding an audio signal, a method and apparatus for inversely transforming an audio signal, and a method and apparatus for adaptively decoding an audio signal. The method of transforming an audio signal includes determining a transform unit into which the audio signal in a time domain is to be transformed into an audio signal in a frequency domain, and transforming the audio signal into an audio signal in the frequency domain according to the determined transform units using a window coefficient other than 0. Accordingly, it is possible to minimize distortion of the audio signal when encoding the audio signal even at a high bit rate while increasing efficiency of compression.

Abstract: An apparatus, method, and medium for distinguishing a vocal sound. The apparatus includes: a framing unit dividing an input signal into frames, each frame having a predetermined length; a pitch extracting unit determining whether each frame is a voiced frame or an unvoiced frame and extracting a pitch contour from the voiced and unvoiced frames; a zero-cross rate calculator respectively calculating a zero-cross rate for each frame; a parameter calculator calculating parameters including a time length ratio of the voiced frame and the unvoiced frame determined by the pitch extracting unit, statistical information of the pitch contour, and spectral characteristics; and a classifier inputting the zero-cross rates and the parameters output from the parameter calculator and determining whether the input signal is a vocal sound.

Abstract: This application discloses A multi-lingual output device for output of transactional information for a given customer, the device that includes a data base for determining what transaction information needs to be outputted, the local language in which the information is to be outputted, and the preferred language of the customer in which the information is to be outputted; and, a local transaction subsystem in communication with said database, wherein said local transaction sub system includes input device receiving means for accepting an input device and output generating means for generating a signal to an output device.

Abstract: A system for enhancing a signal regenerated from an encoded audio signal. The system comprises a decoder arranged to receive the encoded audio signal and produce a decoded audio signal, a feature extraction means arranged to receive at least one of the decoded and encoded audio signal and extract at least one feature from at least one of the decoded and encoded audio signal, a mapping means arranged to map the at least one feature to an enhancement signal and operable to generate and output the enhancement signal, whereby the enhancement signal has a frequency band that is within the decoded audio signal frequency band, and a mixing means arranged to receive the decoded audio signal and the enhancement signal and mix the enhancement signal with the decoded audio signal.

Abstract: A word alignment modeler uses probabilistic learning techniques to train “word-dependent transition models” for use in constructing phrase level Hidden Markov Model (HMM) based word alignment models. As defined herein, “word-dependent transition models” provide a probabilistic model wherein for each source word in training data, a self-transition probability is modeled in combination with a probability of jumping from that particular word to a different word, thereby providing a full transition model for each word in a source phrase. HMM based word alignment models are then used for various word alignment and machine translation tasks. In additional embodiments sparse data problems (i.e., rarely used words) are addressed by using probabilistic learning techniques to estimate word-dependent transition model parameters by maximum a posteriori (MAP) training.

Abstract: An apparatus and method encode audio data, and an apparatus and method decode encoded audio data. An audio data encoding apparatus includes: a scalable encoding unit dividing audio data into a plurality of layers, representing the audio data in predetermined numbers of bits in each of the plurality of layers, and encoding a lower layer prior to encoding an upper layer and an upper bit of each layer prior to encoding a lower bit of each layer; an SBR encoding unit generating spectral band replication (SBR) data that has information with respect to audio data in a frequency band of frequencies equal to or greater than a predetermined frequency among the audio data to be encoded, and encoding the SBR data; and a bitstream production unit generating a bitstream using the encoded SBR data and the encoded audio data corresponding to a predetermined bitrate.

Abstract: Representation-neutral dialogue systems and methods (“RNDS”) are described that include multi-application, multi-device spoken-language dialogue systems based on the information-state update approach. The RNDS includes representation-neutral core components of a dialogue system that provide scripted domain-specific extensions to routines such as dialogue move modeling and reference resolution, easy substitution of specific semantic representations and associated routines, and clean interfaces to external components for language-understanding (i.e., speech-recognition and parsing) and language-generation, and to domain-specific knowledge sources. The RNDS also allows seamless interaction with a community of devices.

Abstract: Methods and apparatus are provided for achieving an arbitrary average data rate for a variable rate coder. One method includes selecting a set (e.g., a pair) of initial composite rates surrounding the arbitrary average data rate. A reallocation fraction is then calculated based on the initial composite rates. The reallocation fraction is used to reassign a number of frames from one component rate of an initial composite rate to another in order to achieve the arbitrary average data rate. Such a method may be configured such that selecting an initial composite rate on one side of (e.g., less than) the arbitrary average data rate implicitly selects the initial composite rate on the other side of the arbitrary average data rate.

Abstract: Embodiments of a feature generation system and process for use in machine learning applications utilizing statistical modeling systems are described. In one embodiment, the feature generation process generates large feature spaces by combining features using logical, arithmetic and/or functional operations. A first set of features in an initial feature space are defined. Some or all of the first set of features are processed using one or more arithmetic, logic, user-defined combinatorial processes, or combinations thereof, to produce additional features. The additional features and at least some of the first set of features are combined to produce an expanded feature space. The expanded feature space is processed through a feature selection and optimization process to produce a model in a statistical modeling system.

Abstract: Embodiments of a progressive feature selection method that selects features in multiple rounds are described. In one embodiment, the progressive feature selection method splits the feature space into tractable sub-spaces such that a feature selection algorithm can be performed on each sub-space. In a merge-split operation, the subset of features that the feature selection algorithm selects from the different sub-spaces are merged into subsequent sets of features. Instead of re-generating the mapping table for each subsequent set from scratch, a new mapping table from the previous round's tables is created by collecting those entries that correspond to the selected features. The feature selection method is then performed again on each of the subsequent feature sets and new features are selected from each of these feature sets. This feature selection-merge-split process is repeated on successively smaller numbers of feature sets until a single final set of features is selected.

Abstract: Audio coding and decoding apparatuses and methods which support fine granularity scalability (FGS) using harmonic information of a high-band audio signal or wideband error audio signal when performing wideband audio coding and decoding, and recording mediums on which the methods are stored. The audio coding method includes detecting harmonics of a high-band audio signal or wideband error audio signal of an input audio signal; determining an order of the detected harmonics; and coding the detected harmonics based on the determined order.

Abstract: An aspect of the invention provides a commercial detection apparatus for detecting commercials that includes a silent detector configured to detect a silent segment based on the strength of the audio signal output in content, and a determination unit configured to determine a sound segment as a commercial if three or more silent segments are detected essentially within a set time span, and if the sound segment is found between two of the three silent segments.