Abstract: A search machine finds and ranks documents in a database based on a set of rules that match characteristics of the database with a natural language query. The system includes a lexicon component which may parse the query and the database into words and word stems. Thereafter, the query and documents may be vectorized such that the elements of the vector correspond to a given word stem, and the value of the element in the vector corresponds to the number of occurrences of the word in the document. The vectorized query is then compared and evaluated against each of the vectorized documents of the database to obtain a ranked list of documents from the database. The user may evaluate the documents found and provide information back to the search machine in order to adjust, for example, the ranking produced by the search machine. In this way, the search machine can fine tune its search and ranking technique to meet the user's specific criteria.

Abstract: A computer system has a notification manager for playing a message to a user by selecting one of a plurality of audio notifications. The method includes the step of setting a priority level for each notification arriving into a queue. The notification is inserted into a position in the queue based upon the priority level of the notification, such that the audio notifications at the queue top have a generally higher priority than audio notifications at the queue bottom. The notification at the top of the queue can be selected if the priority level of the notification is greater than a predetermined gate level. Once a notification is selected, a message corresponding to the selected notification is played to the user.

Abstract: An electronic dictionary that can perform model sentence searches at high speed and can search through a plurality of dictionaries at the same time is obtained by providing dictionary daa having sub-information comprising translations corresponding to word entries, model sentences and idiomatic phrases relating to the word entries, and translations of the model sentences and idiomatic phrases, in which as least the model sentence data of the sub-information are extracted from the dictionary data and stored as sub-dictionary data. A pointer is provided at a position in the dictionary data where the sub-information is embedded, and associates respective sub-information with sub-dictionary data.

Abstract: A method of configuring classes in a natural language understanding (NLU) system. The method can include the steps of assigning a unique value to members of a class in the NLU system. The step of generating possible substrings from the members in the class also can be included. Additionally, for each generated substring having at least one term in common with one of the members in the class, the step of associating with the generated substring the unique value assigned to the member can be included.

Abstract: A system and method for analyzing a speech problem includes the steps of presenting to a student/user a symbol representative of a word and prompting the user to pronounce the word represented by the symbol into a microphone in signal communication with a processor. Next the therapist enters a phonetic representation of the user pronunciation into the processor. It is then automatically determined whether an error exists in the user pronunciation. If an error exists, the error is automatically categorized. The system and method may be adapted for presentation of a single word, a plurality of words having a predetermined feature desired to be tested, a pretest for screening for potential articulation disorders, and an analysis of connected speech with the use of a moving picture to elicit a narrative from the student.

Abstract: A data processing system collects video and audio samples of acceptable speech production. A video camera focuses on a speaker's face and, particularly, articulation visible in the area of the mouth or other body movements associated with speech production. Video files are used to archive acceptable and unacceptable productions. These files may then be used to provide feedback about acceptable and unacceptable ways to produce speech. A speech professional or language teacher may play a model speech production and a subject speech attempt simultaneously to compare articulation, audio analysis, and appearance of articulators. A subject may play a model speech production and record a speech attempt simultaneously to attempt to mimic the appearance of articulators. Image processing may be used to create a mirror image of a video model or a current attempt or both to avoid left-right confusion.

Abstract: A system and method for analyzing a speech problem includes the steps of presenting a symbol representative of a word and prompting the user to pronounce the word represented by the symbol into an audio input device such as a microphone in signal communication with a processor. Next the therapist enters a phonetic representation of the user pronunciation into an operator input and storage device, the phonetic representation subsequently downloaded into the processor. It is then automatically determined whether an error exists in the user pronunciation. If an error exists, the error is automatically categorized.

Abstract: A signal processor for effecting the conversion of an audio data signal from one format to another. The signal processor has a signal converter that can selectively acquire two operative modes, namely a first operative mode and a second operative mode. In the first operative mode, the signal converter transforms the audio data signal from one format to another and releases the converted audio data signal from the output of the signal processor. In the second operative mode, the signal converter is disabled and permits passage of the audio data signal to the output without conversion. The signal processor has a control unit for controlling the transition of the signal converter between operative modes. The control unit enables the signal converter to pass from the first operative mode to the second operative mode when at least one operating condition has been satisfied.

Abstract: A speech recognition feature extractor includes a time-to-frequency domain transformer for generating spectral values in the frequency domain from a speech signal; a partitioning means for generating a first set and an additional set of spectral values in the frequency domain; a first feature generator for generating a first group of speech features using the first set of spectral values; a additional feature generator for generating an additional group of speech features using the additional set of spectral values; the feature generators arranged to operate in parallel, an assembler for assembling an output set of speech features from at least one speech feature from the first group of speech features and at least one speech feature from the additional group of speech features, and an anti-aliasing and sampling rate reduction block, where the first and the additional set of spectral values comprise at least one common spectral value.

Abstract: Apparatus for assessing the performance of telecommunications systems by comparison of a reference signal with the same signal as degraded by the system under test tends to give unreliable results when used to test connectionless packet-switching equipment. This is because variable delay in delivery of the individual packets can result in the test signal and reference signal becoming unsynchronised. This may lead to the apparatus recording a poorer performance than a human tester would have recorded. To overcome this problem, the invention separates the test signal into a series of individual sections, which are synchgronised and analysed separately, and a quality value, aggregated over all the sections, is derived.

Abstract: A transcription method uses a computerized process to prompt a student to produce at least one phoneme orally. Next a correct and at least one incorrect production of the phoneme are displayed. The therapist selects from among the displayed productions based upon the student-produced phoneme. The system includes a processor and display to prompt a student to produce at least one phoneme orally, display a correct and at least one incorrect production of the phoneme. The therapist then uses an input device in signal communication with the processor to select from among the displayed correct and incorrect productions based upon the student-produced phoneme, thus obviating the need for the therapist to enter the incorrect production symbol by symbol, unless it is desired to do so, or unless the actual production is not found among the displayed production selections.

Abstract: A voice operated portable information management system that is substantially language independent and capable of supporting a substantially unlimited vocabulary. The system includes a microphone, speaker, clock and GPS connected to a speech processing system.

Abstract: A method for providing speech therapy includes the steps of selecting a problem speech sound and searching a database that houses a plurality of records. Each record contains a picture and a word associated with the word. Next a set of records is automatically generated from the plurality of records. Each record contains a word specific to the problem speech's sound. At least a portion of each record in the set of records is next automatically presented to a user sequentially on a display device, and the user is prompted to pronounce the displayed word. Finally, the pronunciation of each word is scored. The system includes hardware for carrying out the method, including a processor, display device, input device, and software resident on the processor adapted to access a database of records, present records to the user, and receive a score entered into the input device.

Abstract: In a correction device (3, 4) for correcting a text (ET) recognized by a speech recognition device (2) from a spoken text (GT), marking means (12, 15, 17) are provided for storing marking information (WMI) in memory means (8), which marking information (WMI) flags those parts of the recognized text (ET) and/or of the spoken text (GT) that were reproduced at least once by reproduction means (13) in a synchronous reproduction mode. In this way, parts of the recognized text (ET) and the spoken text (GT) that have already been reproduced once can be acoustically or visually marked, and different parts can be reproduced depending on whether reproduction takes place for the first time or is repeated.

Abstract: A method for recognizing speech commands includes defining a time window for recognizing a speech command. A first confidence value is determined for a first recognition result. If the first confidence value is greater than or equal to a first threshold value, the first recognition result is selected as the speech command, otherwise a second recognition stage is performed where the time window is extended. A second confidence value is determined for the second recognition result. If the second confidence value is greater than or equal to the first threshold value, a command word selected at the second recognition stage is selected as the speech command, otherwise the first and second recognition results are compared to each other to determine a probability that they are substantially the same. If the probability exceeds a predetermined value, the command word selected at the second stage is selected as the speech command.

Abstract: A method and system for providing comfort noise in the non-speech periods in speech communication. The comfort noise is generated based on whether the background noise in the speech input is stationary or non-stationary. If the background noise is non-stationary, a random component is inserted in the comfort noise using a dithering process. If the background noise is stationary, the dithering process is not used.

Abstract: The present invention provides a transform coding method efficient for music signals that is suitable for use in a hybrid codec, whereby a common Linear Predictive (LP) synthesis filter is employed for both speech and music signals. The LP synthesis filter switches between a speech excitation generator and a transform excitation generator, in accordance with the coding of a speech or music signal, respectively. For coding speech signals, the conventional CELP technique may be used, while a novel asymmetrical overlap-add transform technique is applied for coding music signals. In performing the common LP synthesis filtering, interpolation of the LP coefficients is conducted for signals in overlap-add operation regions. The invention enables smooth transitions when the decoder switches between speech and music decoding modes.

Abstract: A method and a system are provided for controlling the coding rates of a multimode coding system with respect to a sequence of input audio signal frames. The method eliminates or minimizes the overflow and underflow of a bit-stream buffer maintained by the coding system for temporarily recording bit-stream data prior to transmission or storage.

Abstract: Systems and methods by which voice/data communications may occur in multiple modes/protocols are disclosed. In particular, systems and methods are provided for multiple native mode/protocol voice and data transmissions and receptions with a computing system having a multi-bus structure, including, for example, a TDM bus and a packet bus, and multi-protocol framing engines. Such systems preferably include subsystem functions such as PBX, voice mail and other telephony functions, LAN hub and data router. In preferred embodiments, a TDM bus and a packet bus are intelligently bridged and managed, thereby enabling such multiple mode/protocol voice and data transmissions to be intelligently managed and controlled with a single, integrated system. A computer or other processor includes a local area network controller, which provides routing and hub(s) for one or more packet networks. The computer also is coupled to a buffer/framer, which serves to frame/deframe data to/from the computer from TDM bus.

Abstract: A method and system for selecting formant trajectories based on input speech and corresponding text data. The input speech is analyzed to obtain formant candidates for the respective time frame. The text data corresponding to the input speech is converted into a sequence of phonemes which are then time aligned such that each phoneme is temporally labeled with a corresponding segment of the input speech. Nominal formant frequencies are assigned to a center timing point of each phoneme and target formant trajectories are generated for each time frame by interpolating the nominal formant frequencies between adjacent phonemes. For each time frame, at least one formant candidate that is closest to the corresponding target formant trajectories is selected according to a minimum cost factor. The selected formant candidates are output for storage or further processing in subsequent speech applications.