Abstract: A method and apparatus to perform speech recognition over a voice channel is described.

Abstract: A speech encoder/decoder for wideband speech with a partitioning of wideband into lowband and highband, convenient coding of the lowband, and LP excited by noise plus some periodicity for the highband. The embedded lowband may be extracted for a lower bit rate decoder.

Abstract: An interactive multimedia book provides hands-on multimedia instruction to the user in response to voiced commands. The book is implemented on a computer system and includes both text and audio/video clips. The interactive multimedia book is accessed by voiced commands and natural language queries as the primary user input. The displayed text is written in a markup language and contains hyperlinks which link the current topic with other related topics. The user may command the book to read the text and, as the text is read by the voice synthesizer, a word which is also a hyperlink will change its attributes upon being spoken. The user will be able to observe or hear this and simply utter the word which is the hyperlink to navigate to the linked topic.

Abstract: Methods for providing users with information in audible form are provided. One such method comprises determining content sources from which information is to be retrieved for responding to a request from a user, retrieving grammar fragments corresponding to the content sources determined, providing a menu format including information for providing the grammar fragments to the user in audible form, and aggregating the grammar fragments retrieved such that the grammar fragments can be provided to the user in audible form in conformance with the menu format. Systems and computer-readable media are also provided.

Abstract: Methods and apparatus are described for effecting a computer transaction using speech as a primary input. The speech is captured using a speech recognition program. A context associated with the captured speech is determined. Where the context has been determined, the computer transaction is built based on the context and at least a portion of the captured speech. A representation of the computer transaction is presented to a human operator for verification. The computer transaction is effected upon verification by the human operator.

Abstract: A method and apparatus are provided for automatically recognizing words of spoken speech using a computer-based speech recognition system according to a dynamic semantic model. In an embodiment, the speech recognition system recognizes speech and generates one or more word strings, each of which is a hypothesis of the speech, and creates and stores a probability value or score for each of the word strings. The word strings are ordered by probability value. The speech recognition system also creates and stores, for each of the word strings, one or more keyword-value pairs that represent semantic elements and semantic values of the semantic elements for the speech that was spoken. One or more dynamic semantic rules are defined that specify how a probability value of a word string should be modified based on information about external conditions, facts, or the environment of the application in relation to the semantic values of that word string.

Abstract: A frame erasure concealment technique for a bitstream-based feature extractor in a speech recognition system particularly suited for use in a wireless communication system operates to “delete” each frame in which an erasure is declared. The deletions thus reduce the length of the observation sequence, but have been found to provide for sufficient speech recognition based on both single word and “string” tests of the deletion technique.

Abstract: A method prepares a functional finite-state transducer (FST) with an epsilon or empty string on the input side for factorization into a bimachine. The method creates a left-deterministic input finite-state automation (FSA) by extracting and left-determinizing the input side of the functional FST. Subsequently, the corresponding sub-paths in the FST are identified for each arc in the left-deterministic FST and aligned.

Abstract: A data processing apparatus for data processing an audio signal includes an input terminal (1) for receiving the audio signal, a 1-bit A/D converter (4) for A/D converting the audio signal to for a bitstream signal, a prediction unit (10) for carrying out a prediction step on the bitstream signal to form a predicted bitstream signal, a signal combination unit (42) for combining the bitstream signal and the predicted bitstream signal to form a residue bitstream signal, and an output terminal (14) for supplying the residual bitstream signal.

Abstract: A computer-aided writing system offers assistance to a user writing in a non-native language, as the user needs help, without requiring the user to divert attention away from the entry task. The writing system provides a user interface (UI) that integrates writing assistance with in-line text entry. When the user is unsure of a word's spelling or whether the word is appropriate, the user may enter a corresponding native word directly in line with the ongoing sentence. An error tolerant spelling tool accepts the native word (even if it is misspelled or mistyped) and derives the most probable non-native word for the given context. The spelling tool consults a bilingual dictionary to determine possible non-native word translation candidates, a non-native language model (e.g.

Abstract: A speech communication apparatus which is used with a microphone being fixed to a predetermined position in the vicinity of the mouth in such a manner as to prevent the transmission of uncomfortable noise such as sneezing, coughing or throat-clearing noise to a partner. There is provided a speech communication apparatus including a speech communication microphone, a speaker and a communication unit for amplifying an output signal from the speech communication microphone, the speech communication apparatus includes the communication unit having an amplifier for amplifying an input signal and outputting the input signal so amplified, and a controller for controlling the gain of the amplifier in response to an excessive input signal, wherein the controller controls the gain of the amplifier such that a reproduced sound of an excessive input signal is reduced to a predetermined level only for a predetermined period of time when the excessive input signal is detected.

Abstract: A method for operating a voice function of a dual-mode mobile communication apparatus including a speaker's voice recognition function and a voice output function of stored information while the mobile communication apparatus is operating in an analog mode is disclosed. The method comprises the step of determining whether a voice function request signal is input or not, switching a vocoder into a digital mode for operating the voice function, and operating the voice function in digital mode.

Abstract: A terminal is designed so that a user who employs numerical keys, allocated for the entry of dots, can enter Braille dot combinations that are used for the input of characters. The input characters may be output as speech for feedback. Further, when a terminal accesses a server, Braille dot combinations can be entered in the above described manner, and speech can be fed back from the server. Furthermore, the server can provide a service for the user in accordance with a character string input at the terminal.

Abstract: A system and method facilitating signal enhancement utilizing mixture models is provided. The invention includes a signal enhancement adaptive system having a speech model, a noise model and a plurality of adaptive filter parameters. The signal enhancement adaptive system employs probabilistic modeling to perform signal enhancement of a plurality of windowed frequency transformed input signals received, for example, for an array of microphones. The signal enhancement adaptive system incorporates information about the statistical structure of speech signals. The signal enhancement adaptive system can be embedded in an overall enhancement system which also includes components of signal windowing and frequency transformation.

Abstract: A method of initiating a data session between a wireless device and an information network over a wireless communication link includes establishing a voice session between the wireless device and a voice recognition application over the wireless communication link. A speech request is conveyed to the voice recognition application for information to be obtained from the information network. The voice recognition application converts the speech request into search criteria. A search of the information network is then initiated using the speech criteria and the voice session is terminated. A data session is then established with the wireless device and the search results are pushed to the wireless device over the wireless communications link.

Abstract: Packet stream is generated by combining a plurality of packets corresponding to style-of-rendition identification information which are selected from among a number of packets usable for producing waveforms corresponding to various styles of rendition. Then, a waveform having characteristics of the style of rendition indicated by the style-of-rendition identification information is produced on the basis of the generated packet stream. The packet stream includes a plurality of packets and time information of the individual packets and controls the pitch, amplitude and shape of the waveform to be produced. By thus combining packets corresponding to the style-of-rendition identification information and producing a waveform on the basis of the packet stream, there can be provided a waveform corresponding to a desired style of rendition in a simplified manner with great facility.

Abstract: Sound signal is received which contains sound characteristics to be represented in musical notation. The characteristics, such as a volume level of the sound signal, are extracted out of the received sound signal, and various parameters for use in subsequent analysis of the sound signal are set in accordance with the extracted characteristics. Also, a desired scale determining condition is set by a user. Pitch of the sound signal is determined using the thus-set parameters. The determined pitch is rounded to any one of scale notes, corresponding to the user-set scale determining condition. Also, a given unit note length is set as a predetermined criterion or reference for determining a note length, and a length of the scale note determined from the received sound signal is determined using the thus-set unit note length as a minimum determination unit, i.e., with an accuracy of the unit note length.

Abstract: A system for suppressing unwanted signals in steerable microphone arrays. The lobes of a steerable microphone array are monitored, to identify lobes having large speech content and low noise content. One of the identified lobes is then used to deliver speech to a speech recognition system, as at a self-service kiosk.

Abstract: A device receives a signal that includes human-interpretable audio information. The device automatically adjusts the volume of the audio information at the received end. The volume control is determined by an automatic volume control gain, which is calculated as a function of the automatic gain control gain, a weighted dynamic range compression gain, and a weighted constant gain.

Abstract: The invention concerns a method of task classification using morphemes which operates on the task objective of a user. The morphemes may be generated by clustering selected ones of the salient sub-morphemes selected from training speech which are semantically and syntactically similar. The method may include detecting morphemes present in the user's input communication, and making task-type classification decisions based on the detected morphemes in the user's input communication. The morphemes may be verbal and/or non-verbal.