Abstract: A wireless communication device is disclosed that accepts recorded audio data from an end-user. The audio data can be in the form of a command requesting user action. Likewise, the audio data can be converted into a text file. The audio data is reduced to a digital file in a format that is supported by the device hardware, such as a .wav, .mp3, .vnf file, or the like. The digital file is sent via secured or unsecured wireless communication to one or more server computers for further processing. In accordance with an important aspect of the invention, the system evaluates the confidence level of the of the speech recognition process. If the confidence level is high, the system automatically builds the application command or creates the text file for transmission to the communication device.

Abstract: Schemes for processing a voice command from a caller are disclosed herein. The schemes may allow a caller to manipulate data using a voice command and designate a device to which to provide the manipulated data. The schemes may be accessed over telecommunication networks. According to one exemplary embodiment, a method of processing a voice command from a caller may include receiving from the caller a voice command to manipulate a set of data. The voice command may be processed using a voice-recognition application. The set of data may be manipulated to obtain a different modified set of data, and the modified set of data may be provided to a device designated by the caller.

Abstract: A method and system for language identification are provided. The system includes a feature set of a plurality of character strings of varying length with associated information. The associated information includes one or more significance scores for a character string for one or more of a plurality of languages. Means are provided for detecting character strings from the feature set within a token from an input text. The system uses a finite-state device and the associated information is provided as glosses at the final nodes of the finite-state device for each character string. The associated information can also include significance scores based on linguistic rules.

Abstract: The method of recognizing speech in an acoustic signal comprises developing acoustic stochastic models of voice units in the form of a set of states of an acoustic signal and using the acoustic models for recognition by a comparison of the signal with predetermined acoustic models obtained via a prior learning process. While developing the acoustic models, the voice units are modeled by means of a first portion of the states independent of adjacent voice units and by means of a second portion of the states dependent on adjacent voice units. The second portion of states dependent on adjacent voice units shares common parameters with a plurality of units sharing same phonemes.

Abstract: A class is to be characterized of a probabilistic classifier or clustering system that includes probabilistic model parameters. For each of a plurality of candidate words or word combinations, divergence of the class from other classes is computed based on one or more probabilistic model parameters profiling the candidate word or word combination. One or more words or word combinations are selected for characterizing the class as those candidate words or word combinations for which the class has substantial computed divergence from the other classes.

Abstract: A training system for text to text application. The training system finds groups of documents, and identifies automatically similar documents in the groups which are similar. The automatically identified documents can then be used for training of the text to text application. The comparison uses reduced size versions of the documents in order to minimize the amount of processing.

Abstract: A set of candidate parallel pages is identified based on trigger words in one or more pages downloaded from a given network location (such as a website). A set of document trees representing each of the candidate pages are aligned to identify translationally parallel content and hyperlinks. The parallel content is further fed into conventional sentence aligner for parallel sentences. And the parallel hyperlinks usually refer to other parallel documents, and lead to a recursive mining of parallel documents.

Abstract: To return a predetermined answer in a predetermined order, even in the event that user utterance contents differ from an original objective.

Abstract: A method for speech quality degradation estimation, a method for degradation measures calculation, and the apparatuses thereof are provided. The first method above estimates the speech quality of a speech signal that is modified by a pitch-synchronous prosody modification method, which comprises the following steps. First, extract at least one source pitchmark from the speech signal, and then maps the source pitchmark(s) to at least one target pitchmark(s). Finally, calculate at least one degradation measure based on the mapping between the source and the target pitchmarks. The degradation measures include several weighted pitch-related functions and duration-related functions, where the weighting functions can be calculated based on the speech signal or the pitchmark(s) mapping mentioned above.

Abstract: A translation tool that facilitates translation of a software product into multiple target human languages without requiring recompilation of any binary deliverables. The translation tool is installed by an end user who wishes to translate the software product into the target human language. The end user does not need any programming knowledge. The translator tool extracts all the strings from various sources in the software product, and displays them on a UI to a translator or exports them to a spreadsheet file. The translator translates all the strings via the UI or by modifying the spreadsheet file and saves the translations. The translator tool uses an MSI utility to package the translated deliverables into an installer. The resulting set of install files are now in the target language and can be deployed without having to recompile any of the binary files (EXEs, DLLs) or other content not requiring translation.

Abstract: The present invention relates to a method of processing a media signal and apparatus therefor. A media signal decoding method according to the present invention includes detecting a channel having a valid value of the multi-channels to be generated and generating the detected channel having the valid value from the downmix signal and the spatial information signal. Accordingly, the present invention is able to reduce a decoding operation quantity by detecting which one of the channels to be generated from a transferred media signal is set to a virtual value and omitting decoding for the generation of the channel set to the virtual value.

Abstract: An apparatus and method for adjusting the friendliness of a synthesized speech and thus generating synthesized speech of various styles in a speech synthesis system are provided. The method includes the steps of defining at least two friendliness levels; storing recorded speech data of sentences, the sentences being made up according to each of the friendliness levels; extracting at least one of prosodic characteristics for each of the friendliness levels from the recorded speech data, said prosodic characteristics including at least one of a sentence-final intonation type, boundary intonation types of intonation phrases in the sentence, and an average value of F0 of the sentence, with respect to the recorded speech data; and generating a prosodic model for each of the friendliness levels by statistically modeling the at least one of the prosodic characteristics.

Abstract: A method for performing conditional maximum entropy modeling includes constructing a conditional maximum entropy model, and incorporating an observation confidence score into the model to reduce an effect due to an uncertain observation.

Abstract: An audio encoder in which two or more preferably different encoders cooperate to generate a joint encoded audio signal. Encoding parameters of the two or more encoders are optimized in response to a measure of distortion of the joint encoded audio signal in accordance with a predetermined criterion. The distortion. measure is preferably a perceptual distortion measure. In one encoder embodiment comprising a sinusoidal and a waveform encoder, a constant total bit rate for each audio frame is distributed between the two encoders so as to minimize perceptual distortion for both the first and the second encoder. Other embodiments consider a set of encoding parameters that is larger than only those that minimize the perceptual distortion of the first encoder. In some embodiments, perceptual distortion may be minimized by optimizing encoding via optimizing entire encoding templates, i.e. a complex set of encoding parameters, for the separate encoders.

Abstract: The present invention is to provide an information processing terminal which can use another expression means to indicate undesirable emotions directly transmitted to a party by a method of directly expressing talking person's emotions in real time, so that the whole image of a calling status can be reviewed afterward and grasped. An information processing terminal 1 including: a voice signal output portion 102 for inputting a voice; an emotion estimation portion 201 for generating parameters of emotions from the inputted voice; and a notification portion 30, 40, 50 for giving notice of various kinds of information, wherein the information processing terminal 1 further includes an emotion specifying portion 203 for specifying an emotion expressed by a distinctive parameter of the generated parameters, and the notification portion 30, 40, 50 gives notice of the specified emotion.

Abstract: Finite-state transducers and weighted finite-state automata may not be determinizable. The twins property can be used to characterize the determinizability of such devices. For a weighted finite-state automaton or transducer, that weighted finite-state automaton or transducer and its inverse are intersected or composed, respectively. The resulting device is checked to determine if it has the cycle-identity property. If not, the original weighted finite-state automaton or transducer is not determinizable. For a weighted or unweighted finite-state transducer, that device is checked to determine if it is functional. If not, that device is not determinizable. That device is then composed with its inverse. The composed device is checked to determine if every edge in the composed device having a cycle-accessible end state meets at least one of a number of conditions. If so, the original device has the twins property. If the original device has the twins property, then it is determinizable.

Abstract: An audio encoding apparatus and the like are disclosed which can improve the sound quality of encoded audio signals even in a case of scalable CELP encoding the audio signals in sections that vary with time. In this apparatus, an enhancement layer extended adaptive codebook generating part (102) generates an extended adaptive codebook (d_enh_ext[i]) from both one frame of core layer drive sound source signals (exc_core[n]) received from a core layer CELP encoding part (101) and past enhancement layer drive sound source signals (exc_enh[n]) received from an adder (106), and further inputs the generated extended adaptive codebook (d_enh_ext[i]) to an enhancement layer extended adaptive codebook (103) for each of sub-frames. That is, the enhancement layer extended adaptive codebook generating part (102) updates the extended adaptive codebook (d_enh_ext[i]) for each of the sub-frames.

Abstract: A method for multifunctional processing of signals in frequency subbands performs subband decomposition and signal processing in two stages. A fullband signal is first splitted, with downsampling, into wide frequency subband (WFS) signals. Processing algorithms not requiring a high frequency resolution but benefiting from downsampling (such as subband acoustic echo cancellation), are applied to the WFS signals by wide subband processing blocks. Processed WFS signals are splitted, preferably without downsampling, into groups of narrow frequency subband (NFS) signals. The NFS signals are processed using processing algorithms (noise suppression, etc.) requiring a higher resolution. Processed NFS signals are synthesized into processed WFS signals, which are recombined into an output signal. Two-stage processing makes it possible to optimize signal processing, while keeping computational costs at low level and avoiding undesirable time delays.

Abstract: Various embodiments are disclosed relating to the real-time monitoring and control for audio devices. An apparatus may include a peripheral audio device configured to operate in an operational mode or a debug mode, the peripheral audio device including an audio enhancement logic configured to include at least one tunable parameter. The apparatus may also include the peripheral audio device being further configured to transmit and receive data via a data channel to allow a debug or test to be performed on the peripheral audio device, while operating in the debug mode, and the at least one tunable parameter to be adjusted.

Abstract: A translated text creator translates a text in which an unknown word is left in an original language representation without being translated, while known words are translated. Translated text created by the translated text creator is displayed. A link setter sets a link for performing a search for the unknown word in a search field of a selected Internet search engine which corresponds to a field of a subject matter of the original text.