Digital Equipment Corporation

Compaq Computer Corporation, Inc.

Massachusetts Institute of Technology

William D Goldenthal

Brian S Eberman

Christopher M Weikart

Jean-Manuel Van Thong

James Robert Glass

Keith Waters

Kenneth F Alden

Debajit Ghosh

In a computerized method a continuous signal is segmented in order to determine statistically stationary units of the signal. The continuous signal is sampled at periodic intervals to produce a timed sequence of digital samples. Fixed numbers of adjacent digital samples are grouped into a plurality of disjoint sets or frames. A statistical distance between adjacent frames is determined. The adjacent sets are merged into a larger set of samples or cluster if the statistical distance is less than a predetermined threshold. In an iterative process, the statistical distance between the adjacent sets are determined, and as long as the distance is less than the predetermined threshold, the sets are iteratively merged to segment the signal into statistically stationary units.

Method and apparatus for clustering-based signal segmentation

In a computerized method for messaging, a Web page includes an embedded tag specifying a message extension program. The Web page is requested from a Web server by a Web browser of a first client computer connected to the Web server computer by a communications network. The extension program corresponding to the message tag is retrieved. A multimedia message is recorded as digital signals in a file of the client computer using the extension program. The file is sent to a second client computer as an electronic mail message.

Browser-based electronic messaging

Speech signals from speakers having known identities are used to create sets of acoustic models. The acoustic models along with their corresponding identities are stored in a memory. A plurality of sets of cohort models that characterize the speech signals are selected from the stored sets of acoustic models, and linked to the set of acoustic models of each identified speaker. During a testing session speech signals produced by an unknown speaker having a claimed identity are processed to generate processed speech signals. The processed speech signals are compared to the set of models of the claimed speaker to produce first scores. The processed speech signals are also compared to the sets cohort models to produce second scores. A subset of scores are dynamically selected from the second scores according to a predetermined criteria. The unknown speaker is validated as the claimed speaker if the difference between the first and a combination of the subset of scores is greater than a predetermined threshold value.

Two-staged cohort selection for speaker verification system

In a computerized method for collecting speech data, Web pages of client computers connected to the Internet are enabled to acquire speech signal and information characterizing the speech. The addresses of the enabled Web pages are stored in a list in a memory of a Web server computer. Based on predetermined criteria and the list, some of the enabled client computers are selected to acquire the speech signal and information. The acquired speech signal and information are transmitted to the server computer to generate, train, and evaluate acoustic-phonetic models.

Speech data collection over the world wide web

In a computerized method, speech signals are analyzed using statistical trajectory modeling to produce time aligned acoustic-phonetic units. There is one acoustic-phonetic unit for each portion of the speech signal determined to be phonetically distinct. The acoustic-phonetic units are translated to corresponding time aligned image units representative of the acoustic-phonetic units. An image including the time aligned image units is displayed. The display of the time aligned image units is synchronized to a replaying of the digitized natural speech signal.

Automated speech alignment for image synthesis

Phonetic recognition is provided by capturing dynamical behavior and statistical dependencies of the acoustic attributes used to represent a subject speech waveform. A segment based framework is employed. Temporal behavior is modelled explicitly by creating dynamic templates, called tracks, of the acoustic attributes used to represent the speech waveform, and by generating the estimation of the acoustic spatio-temporal correlation structure. An error model represents this estimation as the temporal and spatial correlations between the input speech waveform and track generated speech segment. Models incorporating these two components (track and error estimation) are created for both phonetic units and for phonetic transitions. Phonetic contextual influences are accounted for by merging context-dependent tracks and pooling error statistics over the different contexts. This allows for a large number of contextual models without compromising the robustness of the statistical parameter estimates. The transition models also supply contextual information.

Growing community of inventors

Cambridge, MA, United States of America

William D Goldenthal