Abstract: A speaker adaptation system and method for speech models of symbols displays a multi-word symbol to be spoken as a symbol. The supervised adaptation system and method has unsupervised adaptation for multi-word symbols, limited to the set of words associated with each multi-word symbol.

Abstract: A telephone dialing speech recognition method includes determining a location associated with a cellular telephone from geographic indications provided by the cellular telephone, and selects associated search information as a function of the location. Speech based dialers operating in a car environment often have difficulty determining the digits said since some digits have similar sounding names in certain languages. To improve recognition performance, constraints are added to the recognition process, based on the natural constraints of the dialing process. The method utilizes the selected associated search information when recognizing the incoming speech signal. For speech dialing, if the user defines a location where the phone is used, then the “numbering plan” of that country may be used to constrain certain digits. Such constraining of the speech recognizer significantly improves the recognition results.

Abstract: A touchpad pointing device utilized as a pattern input device for a pattern recognition system. The input pattern received from the touchpad pointing device is translated into a start application command. The recognition system forms part of a computer having an operating system and includes a touchpad pointing device, a recognition module and a signal capturer. In addition, a unit for launching an application to run on a computer having an operating system and a pointing device which produces an input pattern in response to movement upon it, is described. The unit includes a recognition module and a launcher. The recognition module attempts to recognize the input pattern as a previously trained launching pattern. The launcher launches the application by providing a launching command which is associated with the launching pattern to the operating device. A security application enables access to the computer upon receipt of an input signature from a user with access privileges.

Abstract: A vocoder based voice recognizer recognizes a spoken word using linear prediction coding (LPC) based, vocoder data without completely reconstructing the voice data. The recognizer generates at least one energy estimate per frame of the vocoder data and searches for word boundaries in the vocoder data using the associated energy estimates. If a word is found, the LPC word parameters are extracted from the vocoder data associated with the word and recognition features are calculated from the extracted LPC word parameters. Finally, the recognition features are matched with previously stored recognition features of other words, thereby to recognize the spoken word.

Abstract: A handwritten pattern and stroke recognition system recognizes an input pattern by comparing it to reference patterns. The input and reference patterns comprise at least one stroke, where a stroke is defined as the series of sample points from pen-down to pen-up. The system of the present invention compares the input and reference patterns stroke by stroke and thus, compares input strokes to reference strokes. The system includes a tangent and stable point extractor, a reference database and a stable-point-based classifier. The extracter determines tangents to at least some of the sample points of the input stroke and determines, which of the sample points between a beginning and end of the input stroke are stable points. The reference database stores tangents of sample points of the reference strokes and stable points of the reference strokes.

Abstract: An instruction input unit for supplying operating instructions to a machine is provided. The unit includes a touchpad, an instruction library and a recognizer. The touchpad receives an input pattern from a user and the instruction library stores a multiplicity of operating instruction patterns. Each operating instruction pattern has an operating instruction associated therewith. The recognizer detects which of the multiplicity of operating instructions patterns the input pattern most closely matches and provides the operating instruction associated with the matched operating instruction pattern to the machine. The recognizer includes a handwriting recognizer for recognizing alphanumeric characters.

Abstract: An instruction input unit for supplying operating instructions to a machine is provided. The unit includes a touchpad, an instruction library and a recognizer. The touchpad receives an input pattern from a user and the instruction library stores a multiplicity of operating instruction patterns. Each operating instruction pattern has an operating instruction associated therewith. The recognizer detects which of the multiplicity of operating instructions patterns the input pattern most closely matches and provides the operating instruction associated with the matched operating instruction pattern to the machine. The recognizer includes a handwriting recognizer for recognizing alphanumeric characters.

Abstract: A pattern recognition method of dynamic time warping of two sequences of feature sets onto each other is provided. The method includes the steps of creating a rectangular graph having the two sequences on its two axes, defining a swath of width r, where r is an odd number, centered about a diagonal line connecting the beginning point at the bottom left of the rectangle to the endpoint at the top right of the rectangle and also defining r−1 lines within the swath. The lines defining the swath are parallel to the diagonal line. Each array element k of an r-sized array is associated with a separate array of the r lines within the swath and for each row of the rectangle, the dynamic time warping method recursively generates new path values for each array element k as a function of the previous value of the array element k and of at least one of the current values of the two neighboring array elements k−1 and k+1 of the array element k.

Abstract: A handwritten pattern recognition system for recognizing an input pattern is provided. The system has a plurality of parameter determining units, each determining the value of a desired parameter for an input pattern to be recognized. The system also includes a pattern match determiner which produces match values for each parameter of the input pattern with its corresponding parameter of each reference parameter. The match determiner also produces an overall match value for each reference pattern. A pattern classifier selects the reference pattern whose parameter set is "closest", by some matching criterion, to that of the input pattern.

Abstract: A vocoder based voice recognizer recognizes a spoken word using linear prediction coding (LPC) based, vocoder data without completely reconstructing the voice data. The recognizer generates at least one energy estimate per frame of the vocoder data and searches for word boundaries in the vocoder data using the associated energy estimates. If a word is found, the LPC word parameters are extracted from the vocoder data associated with the word and recognition features are calculated from the extracted LPC word parameters. Finally, the recognition features are matched with previously stored recognition features of other words, thereby to recognize the spoken word.

Abstract: A method for pattern recognition of indicia composed of alphabetical characters of a language in the form of at least a part of a word. The shape of each character in a word is established by means of a pattern shape recognizer. A probability value designating the relative confidence in the recognition of the character as one of the language characters is assigned to each character. The established shapes of the characters are then sequentially applied to a linguistic recognizer for further processing and the probability values of language structure to each of the language characters is assigned. Then the probability values assigned to the characters in the shape recognizer is combined with the corresponding probability values assigned to the characters in the linguistic recognizer and the path between characters possessing the highest probability values is determined.

Abstract: The system of the present invention has a reference character database, which stores a multiplicity of reference characters, and a feature extractor which extracts shape information for each input character. For each reference character, the database lists shape information and reference line measurements indicating how reference lines pass through it. For example, the reference line measurements can be percentages of a height of a bounding box which bounds each reference and/or input character.In some embodiments, the input characters are collected into a stroke buffer and in other embodiments, they are collected into a line buffer. In all embodiments, the reference line estimation is performed for a series of input characters, whether as a full line of characters or as a collection of strokes.

Abstract: A method of recognizing a character is described which includes the following operations: (a) measuring the angle of the tangent to the character at each of a plurality of sample points on the character, (b) assigning those points having a tangent direction within a first, small predefined angle from the vertical axis with values of Up or Down compass directions, and those points having a tangent direction within the first, small predefined angle from the horizontal axis with values of Left of Right compass directions and (c) selecting special points.