Abstract: Methods and apparatus for implementing a telephone communication system incorporating pager features are described. In accordance with the present invention, when a call to a system subscriber is received, the subscriber is contacted by the telephone switch which receives the call via the use of a pager message, e.g., using a conventional pager system. If the subscriber has a pager module of the present invention, which receives the pager message directed to the subscriber inserted into an enhanced telephone of the present invention, the pager module will automatically initiate a call to the switch where the waiting call is being held or “parked”. The automatically placed call will result in a connection between the switch and called party being established over conventional telephone lines. The connection with the switch is thus established automatically in a manner that is transparent to the subscriber.

Abstract: There is provided a video conferencing system for enabling multiple, remotely located transceivers to communicate with each other, including transceiver that utilize different communication formats or protocols. The video conferencing system includes a telecommunications interface unit and a plurality of transceiver that may connect to the telecommunications unit through telephone or network links. The telecommunications interface unit includes a gateway block corresponding to each transceiver connected to the unit and a switching subsystem for directing communications, particularly video and audio information, between the gateway blocks. The transceivers communicate with their respective gateway blocks using their own communication formats, but the gateway blocks communicate to the switching subsystem using a single common language, such as a standard television signal using the NTSC standard.

Abstract: Methods and apparatus for generating and using both speaker dependent and speaker independent garbage models in speaker dependent speech recognition applications are described. The present invention recognizes that in some speech recognition systems, e.g., systems where multiple speech recognition operations are performed on the same signal, it may be desirable to recognize and treat words or phrases in one part of the speech recognition system as garbage or out of vocabulary utterances with the understanding that the very same words or phrases will be recognized and treated as in-vocabulary by another portion of the system. In accordance with the present invention, in systems where both speaker independent and speaker dependent speech recognition operations are performed independently, e.g.

Abstract: System for distributing multimedia information from a central office to multiple subscriber premises includes an optical backbone for transmitting a plurality of optical bands from the central office into a vicinity of the subscriber premises. A controllable switch, positioned outside the subscriber premises, selects a subset of the plurality of optical bands and blocks others that are not contained within the subset, inducing the subset onto an optical transmission pathway. A receiver positioned within the subscriber premises receives at least one optical band of the subset. The controllable switch, in various embodiments, limits the information received at a subscriber premises to only the information to which the subscriber subscribes. In this manner, the need to scramble signals to prevent theft of services is eliminated, since subscribers only receive the services to which they subscribe.

Abstract: A transmultiplexer is connected between digital facilities and analog facilities for translating information between TDM and FDM formats on the digital and analog facilities, respectively. Information in the TDM format is representative of a plurality of digital telephone signal channels and information in the FDM format is representative of a plurality of analog telephone signal channels. The total capacity of the digital facilities is smaller than the total capacity of the analog facilities. Any one of the digital channels can be selectively coupled to any one of the analog channels. The transmultiplexer is installed at a telephone central office facility as an interface between a cable telephony transmission system and a digital telephone switch. A host computer controls both the transmultiplexer and the digital telephone switch.

Abstract: In a noise reduction communications system, where hands-free operations are utilized, a method and system are described for capturing the ambient noise immediately following speech, and using this sample as the basis for noise cancellation of the speech signal, either in a post-processing or real time processing mode.

Abstract: A telephone wire pair is qualified for digital signal transmission by central office testing. The tip-to-ring capacitance of the wire pair is measured while applying to the wire pair an AC test signal at 600 Hz. The test signal at this frequency minimizes errors due to line inductance and/or a ringing device connected to the wire pair. Reliable measurement of the length of the wire pair is achieved. Flaws in cable sheathing can also be detected based on measurement of tip-to-ground capacitance and/or ring-to-ground capacitance.

Abstract: A method for designing a telecom network includes retrieving data corresponding to a plurality of network node locations, a plurality of subscriber locations and other relevant information such as associated street blocks and distance values. Each network node location and subscriber location combination having a distance value therebetween greater than a maximum distance is initially eliminated. Thereafter, each network node location and subscriber location combination is assigned a priority value, based on constraints such as the proximity between their associated street blocks and/or the maximum capacity of each network node location. The total sum of priority values and total number of subscriber locations are then calculated for those sets of network node locations and corresponding subscriber locations that are selected by the algorithm.

Abstract: Methods and apparatus for producing efficiently sized models suitable for pattern recognition purposes are described. Various embodiments are directed to the automated generation, evaluation, and selection of reduced size models from an initial model having a relatively large number of components, e.g., more components than can be stored for a particular intended application. To achieve model size reduction in an automated iterative manner, expectation maximization (EM) model training techniques are combined, in accordance with the present invention, with model size constraints. In one embodiment, a new reduced size model is generated using a LaGrange multiplier from an input model and input size constraints during each iteration of the size reducing model training process. The reduced size model generated during one iteration of the process serves as the input to the next iteration. Scoring, e.g.

Abstract: There is provided a method for detecting horizontal and vertical lines of an encoded run length image to identify cells of the image that are formed by the lines. Initially, the method comprises the steps of filtering excess details, such as text, from the image and grouping a portion of the image for analysis. A top profile of the image is generated and, then, points along the top profile are determined where substantial changes in the profile position occur. A horizontal section of the top profile which corresponds to a line completely visible from the top is then extracted from between the points. By repeating the above steps, each horizontal line of the image is detected and extracted from the image. In addition, in order to detect the vertical lines of the image, the image is simply rotated 90.degree. and processed with the vertical lines being positioned horizontally.

Abstract: Improved automated synthesis of human audible speech from text is disclosed. Performance enhancement of the underlying text comprehensibility is obtained through prosodic treatment of the synthesized material, improved speaking rate treatment, and improved methods of spelling words or terms for the system user. Prosodic shaping of text sequences appropriate for the discourse in large groupings of text segments, with prosodic boundaries developed to indicate conceptual units within the text groupings, is implemented in a preferred embodiment.

Abstract: A telephone wire pair is qualified for digital signal transmission by central office testing. The tip-to-ring capacitance of the wire pair is measured while applying to the wire pair an AC test signal at 600 Hz. The test signal at this frequency minimizes errors due to line inductance and/or a ringing device connected to the wire pair. Reliable measurement of the length of the wire pair is achieved. Flaws in cable sheathing can also be detected based on measurement of tip-to-ground capacitance and/or ring-to-ground capacitance.

Abstract: A broadband optical fiber network for conveying signals between subscriber locations and a head end. A coherent light source is used at the head end to generate signals for transmission to the subscriber locations. Successive groups of downstream optical transmission segments followed by optical separators or splitters carry the coherent light from the head end to the subscriber locations. Non-coherent light sources providing broadband optical signals are situated at the subscriber locations. Successive groups of upstream optical transmissions segments followed by optical combiners carry the broadband optical light signals to the head end.

Abstract: Methods and apparatus for the generation of speaker dependent garbage models from the very same data used to generate speaker dependent speech recognition models, e.g., word models, are described. The technique involves processing the data included in the speaker dependent speech recognition models to create one or more speaker dependent garbage models. The speaker dependent garbage model generation technique involves what may be described as distorting or morphing of a speaker dependent speech recognition model to generate a speaker dependent garbage model therefrom. One or more speaker dependent speech recognition models may then be combined with the generated speaker dependent garbage model to produce an updated garbage model. The scoring of speaker dependent garbage models is varied in accordance with the present invention as a function of the number of speech recognition models from which the speaker dependent garbage model was created.

Abstract: Improved automated synthesis of human audible speech from text is disclosed. Performance enhancement of the underlying text comprehensibility is obtained through prosodic treatment of the synthesized material, improved speaking rate treatment, and improved methods of spelling words or terms for the system user. Prosodic shaping of text sequences appropriate for the discourse in large groupings of text segments, with prosodic boundaries developed to indicate conceptual units within the text groupings, is implemented in a preferred embodiment.

Abstract: Methods and apparatus for activating telephone services in response to speech are described. A directory including names is maintained for each customer. A speaker dependent speech template and a telephone number for each name, is maintained as part of each customer's directory. Speaker independent speech templates are used for recognizing commands. The present invention has the advantage of permitting a customer to place a call by speaking a person's name which serves as a destination identifier without having to speak an additional command or steering word to place the call. This is achieved by treating the receipt of a spoken name in the absence of a command as an implicit command to place a call. Explicit speaker independent commands are used to invoke features or services other than call placement. Speaker independent and speaker dependent speech recognition are performed on a customer's speech in parallel.

Abstract: Methods and apparatus for automating telecommunications billing operations and reducing operator errors when processing operator assisted telephone calls is disclosed. In accordance with the invention, when an operator assisted telephone call is being serviced, a computer is used to automatically generate default billing charge information, e.g., a default class charge, which is automatically entered unless the operator alters this computer generated information. In addition, the computer provides the operator with messages regarding restrictions which are applicable to the call being processed as well as other useful call processing information. When the operator alters the default class charge, a check is automatically performed to insure that the new class charge is a valid class charge. If the class charge appears to be in error, a message is provided to the operator and the operator is given a chance to alter the class charge information before the call is terminated.

Abstract: What is disclosed is a signal processing system, wherein a method and apparatus identify background noise in a signal containing speech and noise by separating the signal into frames, evaluating energy levels of selected frames, and identifying the frames as noise if certain consistency tests are met, and speech if certain pulsing, monotone, and speech level tests are met, and transition between speech and noise if transition deviation and consistency levels are met.

Abstract: Methods and apparatus for inputting telephone number information into an operator workstation are described. In accordance with the present invention, telephone number information may be entered as either alphabetic information using, e.g., the alphabetic keys on an IBM compatible keyboard, or as numeric information using the numeric keys on the keyboard or on a standard telephone keypad. When telephone number information is input as alphabetic information, the workstation automatically converts the alphabetic character of e.g., a vanity telephone number such as "SKYTELL" into the actual numeric 759 8355 which is then used to complete the call. Error checking is performed as part of the data input and conversion process. If an operator attempts to enter the letters Q or Z into a telephone number field or enters a key other than an alphanumeric the input data is rejected and an error message is sent to the operator.

Abstract: A technique for automatically designing a fraud detection system using a series of machine learning methods. Data mining and constructive induction are combined with more standard machine learning techniques to design methods for detecting fraudulent usage based on profiling customer behavior. Specifically, a rule-learning is used to uncover indicators of fraudulent behavior from a large user database. These indicators are used to create profilers, which then serve as features to the fraud detection system that combines evidence from multiple profilers to generate high-confidence intervention activities when the system is deployed on-line with user data.