Abstract: For verbal command processing, a phoneme module selects recognition phonemes from a phoneme input stream. A user recognition module selects a user profile for a user recognized based on the recognition phonemes. The command module processes a command concurrently identified from the phoneme input stream based on the user profile.

Abstract: A method and an apparatus for classifying spoken content in a teleconference for a follower of the teleconference is disclosed. The method comprises: detecting a topic to which the spoken content belongs; determining a (overall) correlation degree between the follower and the spoken content at least according to a correlation degree between the follower and the topic; and classifying the spoken content according to the (overall) correlation degree between the follower and the spoken content. With the method and the apparatus, the correlation degree between the spoken content in the teleconference and the follower of the teleconference can be determined automatically, and the spoken content can be classified according to the correlation degree, so that the follower can selectively pay attention to some spoken contents during the teleconference, which reduces a burden of the follower and improves conference efficiency.

Abstract: When it is determined that sound data is unrecognizable through a speech recognition process by a first speech recognition unit (3), the same sound data as the sound data inputted to the first speech recognition unit (3) is transmitted to a second server device (60) and a first server device (70). Recognition data is generated which is formed of a character string that is a speech recognition result by the second server device (60) with respect to the sound data, and an acoustic model identifier series generated by a first acoustic model identifier series generation unit (27) of the first server (70) based on the sound data, and the generated recognition data is registered in a first recognition dictionary (3b) of the first speech recognition unit (3).

Abstract: Method and apparatuses for constructing a grammar to describe interactions among a plurality of devices in a network are disclosed. An aspect receives, by a network interface of a device, device capabilities of each of the plurality of devices, generates, by a reduced device list generator of the device, a reduced device list representing groupings of the plurality of devices based on the device capabilities, models, by an interaction sequence modeler of the device, one or more sequences of interactions among the plurality of devices using the reduced device list, constructs, by a grammar construction module of the device, the grammar based on the modeled one or more sequences of interactions, and stores the grammar in a memory of the device.

Abstract: A modeling device comprises a front end which receives enrollment speech data from each target speaker, a reference anchor set generation unit which generates a reference anchor set using the enrollment speech data based on an anchor space, and a voice print generation unit which generates voice prints based on the reference anchor set and the enrollment speech data. By taking the enrollment speech and speaker adaptation technique into account, anchor models with a smaller size can be generated, so reliable and robust speaker recognition with a smaller size reference anchor set is possible.

Abstract: Disclosed herein is an ability to identify a supported language. A current language tag including one or more subtags, each subtag providing language identification information. In response to determining that a language referenced by the current language tag is not a supported language, accessing a stored set of mappings including one or more mappings, each mapping associating a first language tag with a second language tag, determining whether or not a mapping in the set of mappings includes the current language tag as one of the first and second language tags and a new language tag as the other of the first and second language tags, and using the new language tag as the current language tag to identify the supported language.

Abstract: Features are disclosed for automatically identifying a speaker. Artifacts of automatic speech recognition (“ASR”) and/or other automatically determined information may be processed against individual user profiles or models. Scores may be determined reflecting the likelihood that individual users made an utterance. The scores can be based on, e.g., individual components of Gaussian mixture models (“GMMs”) that score best for frames of audio data of an utterance. A user associated with the highest likelihood score for a particular utterance can be identified as the speaker of the utterance. Information regarding the identified user can be provided to components of a spoken language processing system, separate applications, etc.

Abstract: A method and apparatus for voice recognition performed in a voice recognition block comprising a plurality of voice recognition stages. The method includes receiving a first plurality of voice inputs, corresponding to a first phrase, into a first voice recognition stage of the plurality of voice recognition stages, wherein multiple ones of the voice recognition stages includes a plurality of voice recognition modules and multiples ones of the voice recognition stages perform a different type of voice recognition processing, wherein the first voice recognition stage processes the first plurality of voice inputs to generate a first plurality of outputs for receipt by a subsequent voice recognition stage. The method further includes, receiving by each subsequent voice recognition stage a plurality of outputs from a preceding voice recognition stage, wherein a plurality of final outputs is generated by a final voice recognition stage from which to approximate the first phrase.

Abstract: A speech signal is received at an input. At least one electrical value associated with the received speech signal is tracked. A dynamic adjustment of the speech signal is determined. The dynamic adjustment is selected at least in part so as to minimize a distortion and minimize an over-amplification of the speech signal based at least in part upon an analysis of the at least one electrical value. The dynamic adjustment is further selected to obtain a desired output signal characteristic for the speech signal presented at an output. The dynamic adjustment value is applied to the speech signal and the adjusted speech signal is presented at the output. The gain of the signal can also be limited to prevent over-amplification.

Abstract: A conversation user interface enables patients to better understand their healthcare by integrating diagnosis, treatment, medication management, and payment, through a system that uses a virtual assistant to engage in conversation with the patient. The conversation user interface conveys a visual representation of a conversation between the virtual assistant and the patient. An identity of the patient, including preferences and medical records, is maintained throughout all interactions so that each aspect of this integrated system has access to the same information. The conversation user interface presents allows the patient to interact with the virtual assistant using natural language commands to receive information and complete task related to his or her healthcare.

Abstract: Providing a framework for merging automatic speech recognition (ASR) systems having a shared deep neural network (DNN) feature transformation is provided. A received utterance may be evaluated to generate a DNN-derived feature from the top hidden layer of a DNN. The top hidden layer output may then be utilized to generate a network including a bottleneck layer and an output layer. Weights representing a feature dimension reduction may then be extracted between the top hidden layer and the bottleneck layer. Scores may then be generated and combined to merge the ASR systems which share the DNN feature transformation.

Abstract: An audio reproducing method for quickly and correctly extracting extra data, including: receiving a data stream including the extra data including an end marker disposed immediately before main data and data length information, which is length information of the extra data, disposed immediately before the end marker; checking the presence/absence of the end marker; and if the end marker exists, extracting the extra data by using the data length information.

Abstract: A speech simulation system adapted for a user to communicate with others. The system has at least one sensor to sense controlled and coordinated body movement. The system has a computer processor connected to the at least one sensor. The system has a database memory connected to the computer processor. The system has software programming to operate the computer processor. The system has a feedback device connected to the computer processor and directed to the user. The system has an outward audio output device connected to the computer processor to provide sound and a speaker connected to the outward audio output device.

Abstract: A method and device for matching speech to text are disclosed, the method including: receiving a speech input, the mentioned speech input carrying input speech information; obtaining initial text corresponding to the input speech information, and respective pinyin of the initial text; generating at least one approximate pinyin for the initial text based on predetermined pronunciation similarity information; and from a preset mapping relationship table, obtaining additional text corresponding to the respective pinyin of the initial text or to the at least one approximate pinyin of the initial text, wherein the preset mapping relationship table includes a respective record for each word in a word database, including respective pinyin and at least one respective approximate pinyin for said each word, and a respective mapping relation between said respective pinyin, said at least one respective approximate pinyin, and said each word.

Abstract: Systems, methods, and computer-readable storage devices for generating speech using a presentation style specific to a user, and in particular the user's social group. Systems configured according to this disclosure can then use the resulting, personalized, text and/or speech in a spoken dialogue or presentation system to communicate with the user. For example, a system practicing the disclosed method can receive speech from a user, identify the user, and respond to the received speech by applying a personalized natural language generation model. The personalized natural language generation model provides communications which can be specific to the identified user.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer-readable storage medium, and including a method for providing creatives. The method comprises identifying a creative for processing, the creative including a title portion, a body portion and optionally a reference portion. The method further comprises evaluating the body portion and the reference portion including determining when either portion includes one or more candidate words or candidate reference that can be added to the title portion. The method further comprises determining when, if promoted, the candidate one or more words or candidate reference would be redundant to content already included in the title portion. The method further comprises promoting the candidate one or more words or candidate reference into the title portion when no unacceptable redundancy is determined.

Abstract: A language processing application employs a classing function optimized for the underlying production application context for which it is expected to process speech. A combination of class based and word based features generates a classing function optimized for a particular production application, meaning that a language model employing the classing function uses word classes having a high likelihood of accurately predicting word sequences encountered by a language model invoked by the production application. The classing function optimizes word classes by aligning the objective of word classing with the underlying language processing task to be performed by the production application. The classing function is optimized to correspond to usage in the production application context using class-based and word-based features by computing a likelihood of a word in an n-gram and a frequency of a word within a class of the n-gram.

Abstract: A data processor for processing a data stream having audio and video data has an input buffer for buffering the data stream; a data stream analyzer for analyzing the data stream for finding information on a plurality of language-specific content in different languages; a queuing buffer for queuing a plurality of parallel queues, each queue having only language-specific content in the same language; and a feeder for feeding a selected queue in accordance with a language selection signal to a subsequent processing stage, wherein a non-selected queue is not fed by the feeder. Also disclosed are a corresponding method for processing a data stream and a computer-readable digital storage medium.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer-readable storage medium, and including a computer-implemented method for providing creatives. The method comprises identifying, using one or more processors, a creative for processing, the creative including a title portion and a body portion, where the body portion includes a visual uniform resource locator (URL), the visual URL comprising a visual portion and a link to a resource. The method further comprises evaluating the visual URL for inclusion in the title portion, including determining when promotion of the visual URL satisfies one or more promotion criterion, and if so, promoting the visual URL for inclusion in the title portion. The method further comprises providing the creative including the title portion with the promoted visual URL.

Abstract: The present technology provides robust, high quality expansion of the speech within a narrow bandwidth acoustic signal which can overcome or substantially alleviate problems associated with expanding the bandwidth of the noise within the acoustic signal. The present technology carries out a multi-faceted analysis to accurately identify noise within the narrow bandwidth acoustic signal. Noise classification information regarding the noise within the narrow bandwidth acoustic signal is used to determine whether to expand the bandwidth of the narrow bandwidth acoustic signal. By expanding the bandwidth based on the noise classification information, the present technology can expand the speech bandwidth of the narrow bandwidth acoustic signal and prevent or limit the bandwidth expansion of the noise.