Abstract: Portions from time-domain speech segments are extracted. Feature vectors that represent the portions in a vector space are created. The feature vectors incorporate phase information of the portions. A distance between the feature vectors in the vector space is determined. In one aspect, the feature vectors are created by constructing a matrix W from the portions and decomposing the matrix W. In one aspect, decomposing the matrix W comprises extracting global boundary-centric features from the portions. In one aspect, the portions include at least one pitch period. In another aspect, the portions include centered pitch periods.

Abstract: Generic and specific C-to-E binaural cue coding (BCC) schemes are described, including those in which one or more of the input channels are transmitted as unmodified channels that are not downmixed at the BCC encoder and not upmixed at the BCC decoder. The specific BCC schemes described include 5-to-2, 6-to-5, 7-to-5, 6.1-to-5.1, 7.1-to-5.1, and 6.2-to-5.1, where “0.1” indicates a single low-frequency effects (LFE) channel and “0.2” indicates two LFE channels.

Abstract: Method and apparatus for synthesizing multiple localizable file formats into a canonical format. Embodiments may provide a file format-independent localization mechanism that automates the extraction of localizable text content from localizable files of different file formats and generates translation kits formatted according to a canonical format. The generated translation kits may include localizable text content from the localizable files for which translations were not found in a translation database. The generated translation kits may be handed off to translators for translation of the localizable text content in the translation kits. The translated text content in the translation kits may then be imported into the translation database and merged by the localization mechanism into localized versions of the localizable files while preserving the original file structure of the files.

Abstract: Methods, systems, and products are disclosed for creating a voice response grammar in a voice response server that include identifying a user for a presentation, the user having a user grammar, the user grammar including one or more user grammar elements and storing a multiplicity of user grammar elements for the user in a voice response grammar on a voice response server. In typical embodiments, identifying a user for a presentation includes creating a data structure representing a presentation and listing in the data structure at least one user identification. In typical embodiments, each grammar element includes an identifier of a structural element, a key phrase for invoking a presentation action, and a presentation action identifier representing a presentation action.

Abstract: A speaking period required for a voice recognition processing is detected with a simple and robust approach. A speaking period is detected based on an EMG generated when a speaker speaks. A large amplitude is observed in an EMG as muscular activity is caused when a speaker speaks. By observing this, a speaking period can be detected. The EMG can be measured from the speaker's skin via electrodes provided on a mobile-type terminal. Since a mobile-type terminal is usually pressed to the skin for use, affinity between this use form and the present invention is very high.

Abstract: A sound signal encoder for high efficiency encoding of sound signals from a plurality of channels is provided which includes a to-be-correlated object setter (52), to-be-correlated object selector (56) and a variable-length encoder (58). The to-be-correlated object setter (52) sets, on the basis of left-channel frequency information held in a left-channel frequency information holder (50) and right-channel frequency information held in a right-channel frequency information holder (51), index [i] indicating which ones of sine waves on the left channel are to be correlated with, namely, are to be subtracted from, sine waves on the right channel. The to-be-correlated object selector (56) selects a default value read from a storage unit (55) or index [i]-th amplitude information read from a left-channel amplitude information holder (53) as an object to be subtracted from the i-th amplitude information on the right channel according to the index [i].

Abstract: A method and apparatus are provided for automatically forming a grammar. Example text strings are received and N-grams are formed based on the text strings. A rule in the grammar is then generated automatically based in part on the n-grams.

Abstract: A speech recognition grammar creating apparatus, which is capable of eliminating complex labor associated with preparing all rules by taking into account changes of the order of component elements of a speech-recognizing object and possible combinations of component elements including at least one component element that can be omitted. In the speech recognition grammar creating apparatus, an image edit section groups together at least one component element that cannot be omitted and at least one component element that can be omitted, as the speech-recognizing object, into a component element group as an omission-allowed group. An augmented BNF converting section creates the speech recognition grammar by expanding the component element group obtained by the grouping.

Abstract: A method for speech enabling an application can include the step of specifying a speech input within a speech-enabled markup. The speech-enabled markup can also specify an application operation that is to be executed responsive to the detection of the speech input. After the speech input has been defined within the speech-enabled markup, the application can be instantiated. The specified speech input can then be detected and the application operation can be responsively executed in accordance with the specified speech-enabled markup.

Abstract: Some spoken languages can be written, for example, by using a single character to represent a single word. The word can comprise a plurality of phonetic codes. A character from a datastore can be retrieved and compared against an input string which might contain the same phonetic codes.

Abstract: A method of detecting voice activity in a signal smoothes the “voice” or “noise” decision to avoid loss of speech segments. The method is particularly suitable for situations in which the noise level is high. Unlike the prior art method which favors optimizing traffic, this method favors the intelligibility of the signal reproduced after decoding. The signal to be coded is divided into frames. A “voice” or “noise” initial decision is made for each signal frame. The method makes the “voice” decision as soon as there is any increase in the energy of the signal relative to the frame preceding the current frame, even if the increase is slight. The method makes the “noise” decision only if the characteristics of the signal correspond to the characteristics of the noise for at least i consecutive frames (for example i=6). The method has applications in telephony.

Abstract: A normalizer (100, 300) of the accent of accented speech modifies (210, 410) the characteristics of input signals that represent the speech spoken in an individual voice with an accent to form output signals that represent the speech spoken in the same voice but with less or no accent.

Abstract: A device and method for translating language is disclosed. In one embodiment, for example, a method for providing a translated output signal derived from a speech input signal, comprises receiving a speech input signal in a first language, converting the speech input signal into a digital format, comprising a voice model component representing a speech pattern of the speech input signal and a content component representing a content of the speech input signal, translating the content component from the first language into a second language to provide a translated content component; and generating an audible output signal comprising the translated content in an approximation of the speech pattern of the speech input signal.

Abstract: A method and apparatus for identifying a semantic structure from text includes processing the input text to identify self-describing fragments of the input text based on a hierarchical schema defining a domain with at least one top-level node and child nodes. Each identified self-describing fragment includes hierarchical context of a portion of the input text. A semantic structure is provided based on the identified self-describing fragments.

Abstract: A system and method for providing text summarization for use in Web-based content is presented. Text is determined responsive to an executed query. Phrases within the text are identified, and words within the phrases are marked using matches of the words within the phrases with words of the executed query and/or a format rule. Marked words are placed into the summarized text subject to space restrictions. A system and method for building Web-based advertising creatives is also presented. At least one item description responsive to an executed query is identified and a name is extracted. Marked words are placed into the advertising creative subject to space restrictions.

Abstract: A method and system for clustering documents based on generalized sentence patterns of the topics of the documents is provided. A generalized sentence patterns (“GSP”) system identifies a “sentence” that describes the topic of a document. To cluster documents, the GSP system generates a “generalized sentence” form of the sentence that describes the topic of each document. The generalized sentence is an abstraction of the words of the sentence. The GSP system identifies clusters of documents based on the patterns of their generalized sentences. The GSP system clusters documents when the generalized sentence representations of their topics have a similar pattern.

Abstract: Packetized CELP-encoded speech playout with frame truncation during silence and frame expansion method dependent upon voicing classification with voiced frame expansion maintaining phasealignment.

Abstract: Method and system for generating electromyographic or sub-audible signals (“SAWPs”) and for transmitting and recognizing the SAWPs that represent the original words and/or phrases. The SAWPs may be generated in an environment that interferes excessively with normal speech or that requires stealth communications, and may be transmitted using encoded, enciphered or otherwise transformed signals that are less subject to signal distortion or degradation in the ambient environment.

Abstract: Method and apparatus for the on-demand localization of files. Embodiments may provide a file format-independent localization mechanism that automates the extraction of localizable text content from localizable files, the process of generating translations for the extracted localizable text content, and the generation of localized versions of the localizable files including the translations for the extracted localizable content. The localized versions of the files may be automatically generated with correct structure and content, correct file names, and automatically placed in correct file locations by the localization mechanism, and are thus readily available to and locatable by an automated build process for the localized version of the product, thus reducing or eliminating the necessity for human intervention during the localization process.

Abstract: A large amount of human labor is required to transcribe and annotate a training corpus that is needed to create and update models for automatic speech recognition (ASR) and spoken language understanding (SLU). Active learning enables a reduction in the amount of transcribed and annotated data required to train ASR and SLU models. In one aspect of the present invention, an active learning ASR process and active learning SLU process are coupled, thereby enabling further efficiencies to be gained relative to a process that maintains an isolation of data in both the ASR and SLU domains.