Abstract: Embodiments are described for harmonicity estimation, audio classification, pitch determination and noise estimation. Measuring harmonicity of an audio signal includes calculation a log amplitude spectrum of audio signal. A first spectrum is derived by calculating each component of the first spectrum as a sum of components of the log amplitude spectrum on frequencies. In linear frequency scale, the frequencies are odd multiples of the component's frequency of the first spectrum. A second spectrum is derived by calculating each component of the second spectrum as a sum of components of the log amplitude spectrum on frequencies. In linear frequency scale, the frequencies are even multiples of the component's frequency of the second spectrum. A difference spectrum is derived subtracting the first spectrum from the second spectrum. A measure of harmonicity is generated as a monotonically increasing function of the maximum component of the difference spectrum within predetermined frequency range.

Abstract: Method and apparatus for reducing a size of databases required for recorded speech data.

Abstract: The disclosure generally describes computer-implemented methods, software, and systems for self-learning localization services. A computer-implemented method includes: identifying, at a location remote from a first application, a request for localization of a string value associated with the first application from a source language to a target language, sending the string value to a translation request buffer in response to a determination that the localization of the string value in the target language is unavailable, and triggering, in response to satisfaction of at least one heuristic analysis, a translation process of the string value from the source language into the target language where the string value is retrieved from the translation request buffer. In some instances, the location remove from the first application is a centralized localization service accessible by remote requests from a plurality of applications.

Abstract: A “Facet Recommender” creates conversational recommendations for facets of particular conversational topics, and optionally for things associated with those facets, from consumer reviews or other social media content. The Facet Recommender applies a machine-learned facet model and optional sentiment-model, to identify facets associated with spans or segments of the content and to determine neutral, positive, or negative consumer sentiment associated with those facets and, optionally, things associated with those facets. These facets are selected by the facet model from a list or set of manually defined or machine-learned facets for particular conversational topic types. The Facet Recommender then generates new conversational utterances (i.e., short neutral, positive or negative suggestions) about particular facets based on the sentiments associated with those facets. In various implementations, utterances are fit to one or more predefined conversational frameworks.

Abstract: A computerized method for generating and evaluating natural language-generated text involves receiving, in a computer, data input by a user, generating, using a natural language generation technique, multiple instances of text stories based upon both contents of a corpus and the received data; analyzing the multiple instances of text stories as a weighted combination of computed geographic scores, distance scores, information content scores, replacement scores and extra aspect scores, providing a ranked set of the generated text stories to a user, receiving a selection of one of the text stories in the ranked set, and storing the selected story.

Abstract: A computer speech output control method, system, and non-transitory computer readable medium, include a computer speech output control system, including a computer speech output unit configured to output a computer speech, a human speech monitoring circuit configured to determine whether a human conversation is occurring, an interruption priority setting circuit configured to set a priority setting for when the human conversation can be interrupted by the computer speech, and an interruption determining circuit configured to determine whether to cause the computer speech output unit to output the computer speech based on the priority setting and a status of the human conversation.

Abstract: An exemplary computer system configured to user multiple automatic speech recognizers (ASRs) with a plurality of language and acoustic models to increase the accuracy of speech recognition.

Abstract: A text-reading device includes: a visual line direction detection device for a driver; a memory that stores the visual line direction when the driver looks at a display device; a gaze determination device that determines that the driver gazes the display device when a state that the detected visual line direction coincides with the stored visual line direction continues for predetermined time or longer; a voice conversion device that outputs text information of the display device as a voice signal based on an instruction; and a reading control device that inputs the instruction when the driver gazes the display device while the display device displays the text information, and the vehicle starts to move.

Abstract: A frequency domain converter is configured to create a plurality of pieces of frequency domain information by individually converting a plurality of input audio signals, which is acquired at different positions, into frequency domain information. A relative value calculator is configured to calculate a relative value of time frequency components of at least one set of frequency domain information among the plurality of pieces of frequency domain information. A signal determiner is configured to determine whether or not each of the input audio signals includes an audio signal component, which is emitted from a predetermined position, based on whether or not the relative value is included in a range specified and based on a relative threshold value stored in a memory in advance.

Abstract: Computer-implemented techniques can include receiving a selected word in a source language, obtaining one or more parts of speech for the selected word, and for each of the one or more parts-of-speech, obtaining candidate translations of the selected word to a different target language, each candidate translation corresponding to a particular semantic meaning of the selected word. The techniques can include for each semantic meaning of the selected word: obtaining an image corresponding to the semantic meaning of the selected word, and compiling translation information including (i) the semantic meaning, (ii) a corresponding part-of-speech, (iii) the image, and (iv) at least one corresponding candidate translation. The techniques can also include outputting the translation information.

Abstract: From a mixed signal in which a first signal and a second signal are mixed, the second signal is removed at low processing cost and without delay. As a result, an estimated first signal which has low residue of the second signal and low distortion is obtained. An estimated first signal is generated by subtracting a pseudo second signal which is estimated to be mixed in a first mixed signal in which a first signal and a second signal are mixed from the first mixed signal. The pseudo second signal is obtained by a first adaptive filter using a second mixed signal in which the first signal and the second signal are mixed in a different proportion from the first mixed signal. A coefficient update amount of the first adaptive filter is made smaller as compared with a case when the estimated first signal is smaller than the first mixed signal, in case the estimated first signal is larger than the first mixed signal.

Abstract: Mechanisms are provided for performing context based synonym filtering for natural language processing. Content is parsed into one or more conceptual units, wherein each conceptual unit comprises a portion of text of the content that is associated with a single concept. For each conceptual unit, a term in the conceptual unit is identified that has a synonym to be utilized during natural language processing of the content. A first measure of relatedness of the term to at least one other term in the conceptual unit is determined. A second measure of relatedness of the synonym of the term to the at least one other term in the conceptual unit is determined. A determination whether or not to utilize the synonym when performing natural language processing on the conceptual unit is made based on the first and second measures of relatedness and natural language processing on the content is performed accordingly.

Abstract: A method of utilizing a speech assistant, the speech assistant designed to provide a voice input and speech output capability, the method comprising, enabling the use of the speech assistant for communication with a user, and terminating the speech assistant when the communication is complete. The method further comprises receiving a notification from a native application associated with the communication, and activating a sub-portion of the speech assistant, to enable outputting of the notification using speech output, thereby enabling the use of speech output for periodic announcements without enabling the speech assistant.

Abstract: Methods described herein provide functionality for automatic speech recognition (ASR). One such embodiment performs speech recognition using received speech recognition result candidates, where the received candidates were generated by performing Statistical Language Model (SLM) based speech recognition on one or more frames of audio data. In turn, such an embodiment transmits results of the speech recognition, performed using the received speech recognition result candidates, to a user device via a communications network. Results of the speech recognition are available with lower latency than pure cloud based ASR solutions.

Abstract: Features are disclosed for performing functions in response to user requests based on contextual data regarding prior user requests. Users may engage in conversations with a computing device in order to initiate some function or obtain some information. A dialog manager may manage the conversations and store contextual data regarding one or more of the conversations. Processing and responding to subsequent conversations may benefit from the previously stored contextual data by, e.g., reducing the amount of information that a user must provide if the user has already provided the information in the context of a prior conversation. Additional information associated with performing functions responsive to user requests may be shared among applications, further improving efficiency and enhancing the user experience.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for recognizing speech using neural networks. One of the methods includes receiving an audio input; processing the audio input using an acoustic model to generate a respective phoneme score for each of a plurality of phoneme labels; processing one or more of the phoneme scores using an inverse pronunciation model to generate a respective grapheme score for each of a plurality of grapheme labels; and processing one or more of the grapheme scores using a language model to generate a respective text label score for each of a plurality of text labels.

Abstract: Methods and systems are provided for separating signal-correlated and signal-uncorrelated error components in quantization noise. Such separation leads to a generalization of the conventional rate-distortion optimization problem. For the commonly used assumption of a Gaussian process, a quantizer according to this principle is implemented in a straightforward manner using a dithered quantizer and appropriate pre-filters and post-filters. If the penalization of the signal-uncorrelated error component is increased over that of the signal-correlated error component, then the pre-filter emphasizes the signal spectrum more, reducing the differential entropy rate of the pre-filtered signal. Accordingly, the signal-uncorrelated noise is reduced for a given rate.

Abstract: The present document relates to methods and systems for music information retrieval (MIR). In particular, the present document relates to methods and systems for extracting a chroma vector from an audio signal. A method (900) for determining a chroma vector (100) for a block of samples of an audio signal (301) is described. The method (900) comprises receiving (901) a corresponding block of frequency coefficients derived from the block of samples of the audio signal (301) from a core encoder (412) of a spectral band replication based audio encoder (410) adapted to generate an encoded bitstream (305) of the audio signal (301) from the block of frequency coefficients; and determining (904) the chroma vector (100) for the block of samples of the audio signal (301) based on the received block of frequency coefficients.

Abstract: A method, system, and non-transitory compute readable medium determining and discerning items with multiple meanings in a sequence of items including producing a distributed representation for each item of the sequence of items including a word vector and a context vector, partitioning the sequence of items into classes, for an item using a representative word vector of each class, calculating a cosine distance between the word vector of said item and the class representative vector, and producing a new sequence of items by modifying the distributed representation in the producing by replacing each occurrence of an item depending on the cosine distance calculated by the calculating.

Abstract: There is provided an information processing apparatus including an information acquiring unit that acquires information to identify an editing point of content including a voice, on the basis of language analysis of the content, and an information output unit that outputs the acquired information.