565 results about "Word recognition" patented technology

Speech recognition models are dynamically re-configurable based on user information, application information, background information such as background noise and transducer information such as transducer response characteristics to provide users with alternate input modes to keyboard text entry. Word recognition lattices are generated for each data field of an application and dynamically concatenated into a single word recognition lattice. A language model is applied to the concatenated word recognition lattice to determine the relationships between the word recognition lattices and repeated until the generated word recognition lattices are acceptable or differ from a predetermined value only by a threshold amount. These techniques of dynamic re-configurable speech recognition provide for deployment of speech recognition on small devices such as mobile phones and personal digital assistants as well environments such as office, home or vehicle while maintaining the accuracy of the speech recognition.

A method of recognizing a handwritten word of cursive script comprising providing a template of previously classified words, and optically reading a handwritten word so as to form an image representation thereof comprising a bit map of pixels. The external pixel contour of the bit map is extracted and the vertical peak and minima pixel extrema on upper and lower zones respectively of this external contour are detected. Feature vectors of the vertical peak and minima pixel extrema are determined and compared to the template so as to generate a match between the handwritten word and a previously classified word. A method for classifying an image representation of a handwritten word of cursive script is also provided. Also provided is an apparatus for recognizing a handwritten word of cursive script. This apparatus comprises a template of previously classified words, a reader for optically reading a handwritten word, and a controller being linked thereto for generating a match between the handwritten word and the previously classified words. Two algorithms are provided for respectively correcting the skew and slant of a word image.

One-to-many comparisons of callers' words and / or voice prints with known words and / or voice prints to identify any substantial matches between them. When a customer communicates with a particular entity, such as a customer service center, the system makes a recording of the real-time call including both the customer's and agent's voices. The system segments the recording to extract different words, such as words of anger. The system may also segment at least a portion of the customer's voice to create a tone profile, and it formats the segmented words and tone profiles for network transmission to a server. The server compares the customer's words and / or tone profiles with multiple known words and / or tone profiles stored on a database to determine any substantial matches. The identification of any matches may be used for a variety of purposes, such as providing representative feedback or customer follow-up.

Disclosed embodiments of the invention provide automated global optimization methods and systems of OCR, tailored to each document being digitized. A document-specific database is created from an OCR scan of a document of interest, which contains an exhaustive listing of words in the document. Images of each word, taken from all the fonts encountered, are entered into the database and mapped to a corresponding textual representation. After entry of a first instance of an image of a word written in a particular font, each new occurrence of the word in that font can be quickly recognized by image processing techniques. The disclosed methods and systems may be used in conjunction with adaptive character recognition training and word recognition training of the OCR engines.

One-to-many comparisons of callers' words and / or voice prints with known words and / or voice prints to identify any substantial matches between them. When a customer communicates with a particular entity, such as a customer service center, the system makes a recording of the real-time call including both the customer's and agent's voices. The system segments the recording to extract different words, such as words of anger. The system may also segment at least a portion of the customer's voice to create a tone profile, and it formats the segmented words and tone profiles for network transmission to a server. The server compares the customer's words and / or tone profiles with multiple known words and / or tone profiles stored on a database to determine any substantial matches. The identification of any matches may be used for a variety of purposes, such as providing representative feedback or customer follow-up.

The invention discloses a microblog user interest recognizing method based on text mining, and belongs to the field of text mining and natural language processing. The method includes the steps of collecting the newest topical microblog text data of a microblog text set and microblog text data of a designated user, standardizing the collected microblog text data, recognizing the newest microblog words and renewing a new word dictionary for the standardized topical microblog text data through the microblog new word recognition method, conducting Chinese character word separation on the standardized microblog text data of the designated user through the new word dictionary word separation method to achieve text vector expression, clustering the microblog text data, expressed through text vectors, of the designated user, recombining original microblog text data, extracting new text set features through a topic model, presetting topic dictionaries, calculating the weight of each topic dictionary based on the new text set features to obtain the final topic, and enabling the final topic to serve as the microblog user interest recognition, thereby improving accuracy of feature extraction.

Speech recognition models are dynamically re-configurable based on user information, application information, background information such as background noise and transducer information such as transducer response characteristics to provide users with alternate input modes to keyboard text entry. Word recognition lattices are generated for each data field of an application and dynamically concatenated into a single word recognition lattice. A language model is applied to the concatenated word recognition lattice to determine the relationships between the word recognition lattices and repeated until the generated word recognition lattices are acceptable or differ from a predetermined value only by a threshold amount. These techniques of dynamic re-configurable speech recognition provide for deployment of speech recognition on small devices such as mobile phones and personal digital assistants as well environments such as office, home or vehicle while maintaining the accuracy of the speech recognition.

A global optimization framework for optical character recognition (OCR) of low-resolution photographed documents that combines a binarization-type process, segmentation, and recognition into a single process. The framework includes a machine learning approach trained on a large amount of data. A convolutional neural network can be employed to compute a classification function at multiple positions and take grey-level input which eliminates binarization. The framework utilizes preprocessing, layout analysis, character recognition, and word recognition to output high recognition rates. The framework also employs dynamic programming and language models to arrive at the desired output.

A computer-based detection (e.g., speech recognition) system combines a word decoder and subword decoder to detect words (or phrases) in a spoken input provided by a user into a speaker connected to the detection system. The word decoder detects words by comparing an input pattern (e.g., of hypothetical word matches) to reference patterns (e.g., words). The subword decoder compares an input pattern (e.g., hypothetical words matches based on subword or phoneme recognition) to reference patterns (e.g., words) based on a word pronunciation distance measure that indicates how close each input pattern is to matching each reference pattern. The subword decoder sorts the source set of reference patterns based on a closeness of each reference pattern to correctly matching the input pattern based on generated pattern comparisons. The word decoder and subword decoder each provide an N-best list of hypothetical matches to the spoken input. A list fusion module of the detection system selectively combines the two N-best lists to produce a final or combined N-best list. The final or combined list has a predefined number of matches.

The invention provides a continuous voice recognition method based on a deep long and short term memory recurrent neural network. According to the method, a noisy voice signal and an original pure voice signal are used as training samples, two deep long and short term memory recurrent neural network modules with the same structure are established, the difference between each deep long and short term memory layer of one module and the corresponding deep long and short term memory layer of the other module is obtained through cross entropy calculation, a cross entropy parameter is updated through a linear circulation projection layer, and a deep long and short term memory recurrent neural network acoustic model robust to environmental noise is finally obtained. By the adoption of the method, by establishing the deep long and short term memory recurrent neural network acoustic model, the voice recognition rate of the noisy voice signal is improved, the problem that because the scale of deep neutral network parameters is large, most of calculation work needs to be completed on a GPU is avoided, and the method has the advantages that the calculation complexity is low, and the convergence rate is high. The continuous voice recognition method based on the deep long and short term memory recurrent neural network can be widely applied to the multiple machine learning fields, such as speaker recognition, key word recognition and human-machine interaction, involving voice recognition.

Alphabetic filtering of the speech recognition of words uses a key press to indicate a desired character in an alphabetic filter string, where each key press represents two or more letters. The key presses can be disambiguated by recognizing a key-disambiguation utterance in association with a given key press. A user can select a desired recognition candidate from a choice list produced by such filtered word recognition. Ambiguous alphabetic filtering can be performed iteratively in response to the addition of successive ambiguous key presses. A user can select to re-recognize the utterance using filtering based on ambiguous key input after seeing the results of recognition without such filtering. Unambiguous alphabetic filtering can be performed by using multiple presses of an ambiguous key to disambiguate which letter is intended. A user can select between entering text by either large vocabulary speech recognition or by spelling text by pressing phone keys.

The invention proposes a voice wakeup method and a voice wakeup device based on artificial intelligence. The voice wakeup method comprises the following steps: acquiring pronunciation information corresponding to a custom wakeup word; acquiring approximate pronunciation information corresponding to the pronunciation information; and building a wakeup word recognition network according to a preset spam word list, the pronunciation information and the approximate pronunciation information so as to recognize voice input by a user according to the wakeup word recognition network and determine whether or not need to perform a wakeup operation according to the recognition result. According to the embodiments of the invention, a wakeup word recognition network can be built dynamically for different custom wakeup, so that the accuracy of wakeup is enhanced effectively, the rate of false alarm is reduced, and the wakeup efficiency is improved. The device has smaller memory footprint and low power consumption.

Disclosed is a method for recognizing a voice and converting it into a text, and more particularly, a method for preferentially voice-recognizing a word starting with a character inputted through a keypad before voice is inputted and converting it into a text, and a mobile terminal implementing such method. The mobile terminal includes a keypad to receive an input of a specific character from a user; a controller to recognize voice input from the user based on words starting with the specific character, and to convert the recognized voice into text; and a display to display the text under a control of the controller.