131 results about "Speech classification" patented technology

There is provided a method of selecting a pitch lag value from a plurality of pitch lag candidates for coding a speech signal. The method comprises identifying the plurality of pitch lag candidates from a frame of the speech signal using correlation; classifying the speech signal to obtain a voice classification; determining whether one or more of the plurality of pitch lag candidates are in a temporal neighborhood of one or more previous pitch lag values; favoring the one or more of the plurality of pitch lag candidates determined to be in the temporal neighborhood of the one or more previous pitch lag values, by adaptive weighting, over other ones of the plurality of pitch lag candidates; and selecting the pitch lag value based on the voice classification and the one or more of the plurality of pitch lag candidates favored by the adaptive weighting.

The invention discloses a speech classification method based on a deep neural network, and aims at solving different speech classification problems through a unified algorithm model. The method of theinvention includes the steps of S1, converting a speech into a corresponding spectrogram; segmenting the complete spectrogram along the frequency domain into blocks to obtain a local frequency domaininformation set; S2, taking the complete and local frequency domain information as inputs of a model respectively, and based on the different inputs, the convolutional neural network being capable ofextracting local and global features; S3, using an attention mechanism to fuse global and local feature expressions to form a final feature expression; S4, using tagged data to train the network by gradient descent and back propagation algorithms; and S5, using the trained parameters for an untagged speech and taking the classification of highest probability that the model outputs as a predictionresult. The method of the invention realizes a unified algorithm model for different speech classification problems, and improves the accuracy on multiple speech classification problems.

The invention relates to a tinnitus hearing aid which comprises a tinnitus masking module and a hearing compensation module, and is characterized in that the tinnitus hearing aid further comprises a control module; the control module comprises a pre-setting unit, a selecting unit and a voice classification device; the pre-setting unit is used for selecting any one function mode in tinnitus masking, hearing compensation and tinnitus masking and hearing compensation, and inputting related preset parameters; the voice classification device is used for sequentially performing pre-treatment, feature extraction and classification according to received signals, and finally determining a sound field environment; a signal processing algorithm is selected and the parameters in the signal processingalgorithm are adjusted according to the function mode selected by the pre-setting unit and the determined sound field environment, and optimal combination is further performed; and the selecting unitis used for opening the voice classification device and working according to the function mode selected by the pre-setting unit.

The invention relates to a navigation apparatus and a voice broadcast method thereof, the method comprises the steps: storing voice classification information and priority setting information corresponding to voice types; generating new voice; continuing the following step in case of judging that the voice is currently broadcasted; and in case of judging that the priority of the type which the new voice belongs to is higher than that of the type which the voice currently broadcasted belongs to, stopping broadcasting the voice which is currently broadcasted and broadcasting the new voice. According to the technical solution of the invention, the situation in which two or more than two voices are broadcasted simultaneously is prevented at any moment, thus avoiding the circumstance that disordered voices, which are resulted from simultaneous broadcast of a plurality of voices, cause inconvenience to users.

The present invention deals with a method and system for classifying at least one speech-user-utterance in a speech classification system to one of a plurality of pre-defined class types. The method comprises transcribing automatically the at least one speech-user-utterance to obtain at least one automatic-transcribed-text and estimating an estimated-transcribed-text corresponding to the at least one automatic-transcribed-text. The method further comprises classifying the at least one speech-user-utterance based on the estimated-transcribed-text. The estimated-transcribed-text is estimated based on at least one statistical model.

A method for the detection of noise and speech segments in a digital audio input signal, the input signal being divided into a plurality of frames including a first stage in which a first classification of a frame as noise is performed if the mean energy value for this frame and the previous N frames is not greater than a first energy threshold, N>1, a second stage in which for each frame that has not been classified as noise in the first stage it is decided if the frame is classified as noise or as speech based on combining at least a first criterion of spectral similarity of the frame with acoustic noise and speech models, a second criterion of analysis of the energy of the frame and a third criterion of duration, and of using a state machine for detecting the beginning of a segment as an accumulation of a determined number of consecutive frames with acoustic similarity greater than a first threshold and for detecting the end of the segment; a third stage in which the classification as speech or as noise of the signal frames carried out in the second stage is reviewed using criteria of duration.

The embodiment of the invention, which relates to the technical field of the computer, provides a voice recognition method and apparatus, so that a problem of a low voice classification distinguishing degree during the voice classification process according to the existing voice recognition technology can be solved. The method comprises: at least two voice features of a to-be-recognized voice are extracted; on the basis of a multi-layer restricted Boltzmann machine (RBM), each of the at least two voice features is trained to obtain a depth voice feature corresponding to each voice feature; feature fusion is carried out on the depth voice feature corresponding to each voice feature, thereby obtaining a depth voice feature of the to-be-recognized voice; and the depth voice feature of the to-be-recognized voice is inputted into a classifier and is classified, so that a voice type of the to-be-recognized voice is obtained. The method and apparatus are applied to voice recognition.

A safety system (10) for a host vehicle (12) includes a voice sensor (38) that detects voice signals from a vehicle occupant. An occupant classifier (30) determines a state of mind of the vehicle occupant. A controller (13) performs a safety countermeasure in response to the state of mind of the occupant including the transmission of the state of mind to a target vehicle. A vehicle voice control system (24) includes the voice sensor (38), which detects voice signals from the vehicle occupant. A speech classifier (32) monitors a speech characteristic, including a vehicle occupant identifiable and associated speech characteristic, in response to the voice signals and associates the voice signals with a vehicle related task in response thereto. A controller (13) performs the vehicle related task in response to the voice signals and the association.

The invention provides a speech recognition method which comprises the following steps of: receiving speech information and a recognition result type transmitted from a client; looking for a speech classification lexicon corresponding to the recognition result type from more than one preset speech classification lexicons, thus obtaining the recognition result from the found speech classification lexicon according to the speech information; and feeding back recognition results which are sequenced according to similarities from high to low to the client. The invention further provides a speech recognition system. According to the technical scheme provided by the invention, the speech recognition rate of application with the speech recognition function is improved.

An object of the present invention is to allow classification of sequentially input speech signals with good accuracy based on similarity of speakers and environments by using a realistic memory use amount, a realistic processing speed, and an on-line operation. A speech classification probability calculation means 103 calculates a probability (probability of classification into each cluster) that a latest one of the speech signals (speech data) belongs to each cluster based on a generative model which is a probability model. A parameter updating means 107 successively estimates parameters that define the generative model based on the probability of classification of the speech data into each cluster calculated by the speech classification probability calculation means 103 (in FIG. 1).

The invention relates to a method and a system for graded measurement of voice. The method comprises the following steps: carrying out voice recognition for a received voice signal, and acquiring a voice feature sequence of state alignment according to a reference text and a reference model; correcting a distribution parameter of the reference model according to the voice feature sequence of state alignment, and generating a voice template vector based on the reference model for the voice signal; using a support vector machine classification decision tree to carry out classification decision for the voice template vector, and then, obtaining the classification grade mapped by the voice template vector. A support vector machine is provided by the invention to build a model for a language classification boundary, and the model is applied to language learning with the following steps: fetching the voice feature of the received voice signal, and carrying out state alignment between the received voice signal and the reference model; correcting the distribution parameter of the reference model and generating a corresponding voice template vector; and using the support vector machine classification decision tree to decide the voice template vector, thus, the complexity of the classification decision of voice is effectively reduced, and the accuracy of the graded measurement of voice is improved.

A device and method are provided for real-time classification of covert speech. The device comprises a plurality of sensors for capturing real-time bio-signal data for brain monitoring in response to mental tasks delivered to a user, and a brain computer interface with memory storing instructions to configure a processor to perform a method of real-time classification of covert speech. The method comprises capturing real-time bio-signal data for brain monitoring in response to mental tasks delivered to a user, pre-processing the raw bio-signal data, extracting a vector of features from the raw bio-signal data, selecting features from the vector of features, building classification model to generate classified covert speech data using the selected features, and controlling a display device with visual elements based on the classified covert speech data.

The invention discloses a speech recognition method, a device, equipment and a computer-readable storage medium. The method comprises the following steps: extracting preselected feature vectors of real speech in a pre-established corpus; extracting preselected feature vectors of real speech in the pre-established corpus; extracting preselected feature vectors of real speech in the pre-establishedcorpus; extracting preselected feature vectors of real speech in the pre-established corpus; according to the preset speech synthesis algorithm and the text information corresponding to the real speech in the corpus, extracting the forged speech is synthesized and the preselected feature vector of the forged speech; training a true/false speech classifier by using a preselected feature vector of the real speech and a preselected feature vector of the forged speech; recognizing the speech to be detected by the trained true-false speech separator so as to judge whether the speech to be detectedis forged speech or not. The method, the device, equipment and the computer-readable storage medium provided by the invention can accurately and efficiently recognize the counterfeit speech in the actual scene and effectively prevent the pseudo sound attack.

The invention discloses a control system and method of the intelligent speech classification and recognition of garbage based on the internet of things, and relates to the technical field of garbage treatment. The control system comprises a garbage dumping box, wherein a plurality of garbage cans are arranged in the garbage dumping box and are used for containing different types of garbage, garbage dumping ports are respectively formed in positions, which correspond to the garbage cans, of the side wall of the garbage dumping box, and electric control doors and infrared sensors which are usedfor detecting whether the garbage cans are filled with garbage are arranged at the garbage dumping ports; a solar panel, a power supply module, a control module, a voice input module, a voice acquisition and recognition module and a voice playing module are arranged on the garbage dumping box, the solar panel is electrically connected with the power supply module, the power supply module is used for supplying power to the control module, the voice input module, the voice acquisition and recognition module, the voice playback module, the electric control doors and the infrared sensors are all electrically connected with the control module, and the control module is connected with a garbage treatment center control station in a wireless communication mode. According to the control system, the internet of things is used, so that the garbage treatment is intelligent.

A method for determining emotional status of a group of N people comprising steps of: receiving an audio data of at least one person from the group of N people by a computer readable medium (CRM); determining emotional status of that person according to the received audio data using the CRM; repeating the steps of receiving and determining for a sub-group of M people selected from the group N people; M is an integer between 1 and N; the method additionally comprising step of determining emotional status of the group by applying a statistical function on plurality of the emotional status collected in the steps mentioned above.

The invention provides an emotion identification method and system for old people. The method comprises the steps of obtaining voice information and image information; processing the voice information to obtain a Mel-frequency cepstral coefficient; processing the image information to obtain expression images; according to the Mel-frequency cepstral coefficient and a preset voice classification rule, classifying the obtained voice information to obtain voice emotion types; and according to a preset expression classification network model, extracting classification features of the expression images, and obtaining image emotion types according to the classification features. The voice emotions and the image emotions of a user are identified according to the obtained voice information and image information, so that current emotions of the user are known from two aspects of voices and expressions, and the emotions of the user are identified more objectively and accurately; the method and the system are of important significance for elderly users; and the emotions, especially, the negative emotions of old people are discovered in time, so that the body conditions of old people can be known in time and old people can be taken care of timely and accurately.

The invention discloses an early education robot speech interaction education system and method and belongs to the robot speech interaction field. The device includes a monitoring unit, a classification unit and a processing unit; the monitoring unit monitors an external environment in real time and obtains speech sent by the external environment; the classification unit compares the speech with apreset speech database, wherein the speech can be wake-up words, dangerous sound, voices of strangers, and the speech of children; the processing unit performs processing according to a classification result; if the speech is the wake-up words, speech subsequently emitted by a person who emits the wake-up words is stored, and a replay is given according to the speech; if the speech is the dangerous sound or the voices of strangers, alarm information is sent to the terminal of a parent; and if the speech is the speech of children, the speech of children is inputted into a preset speech emotionmodel, calculation is performed, and the speech of children is divided into positive emotion speech or negative emotion speech, and if the speech of children is the positive emotion speech, a prompting message is sent, and if the speech of children is the negative emotion speech, and comforting speech is played, and a prompting message is sent.

A method for robust speech classification in speech coding and, in particular, for robust classification in the presence of background noise is herein provided. A noise-free set of parameters is derived, thereby reducing the adverse effects of background noise on the classification process. The speech signal is identified as speech or non-speech. A set of basic parameters is derived for the speech frame, then the noise component of the parameters is estimated and removed. If the frame is non-speech, the noise estimations are updated. All the parameters are then compared against a predetermined set of thresholds. Because the background noise has been removed from the parameters, the set of thresholds is largely unaffected by any changes in the noise. The frame is classified into any number of classes, thereby emphasizing the perceptually important features by performing perceptual matching rather than waveform matching.

The present application relates to the field of speech recognition, and provides a method, apparatus, computer device and storage medium for recognizing speech categories. The method comprises: acquiring first speech information to be recognized and converting the first speech information into a first language spectrum map; inputting the first language spectrum map into a preset speech classification model to obtain a classification result of the first language spectrum map, and taking the classification result as a class of the first speech information; wherein, the speech classification model is trained based on a depth convolution neural network by using a language map of a known emotion class or a character class; The speech recognition method, apparatus, computer device and storage medium provided in the present application are convenient to improve the effect of emotion and character classification in speech information.

A system and method of regulating speech output by a text-to-speech (TTS) system includes: evaluating speech that has been converted from text using an initial speech quality test before presentation to a user; applying a classification test to the evaluated speech if the evaluated speech falls below a threshold based on the initial speech quality test; generating an abnormal speech classification for the evaluated speech; and applying a corrective action to the evaluated speech based on the abnormal speech classification.