Engineering Papers⌕ Search

Engineering topics

McClanahan, Richard

Publications and source records attributed to McClanahan, Richard.

3 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2018

Journey analysis system and method

The various technologies presented herein relate to determining a location of an individual, or an object, when a GPS may be unavailable/occluded. Readings from a plurality of sensors on a mobile device can be generating a plurality of data (e.g., speed, motion, temperature, biometric, etc.) which can be combined to generate a probability of a user of the mobile device being at a particular location, or not. Sensor readings can be weighted to enable sensors having a large effect on a degree of confidence of the probability of the user being determined to be at a location being given priority (e.g., battery priority) versus those sensors which have lesser effect on the degree of confidence. With the user determined to be at a location, information, a notification, an instruction, etc., can be presented to the user, via the mobile device.

47 OTHER INSTRUMENTATION↗
DOE OSTI2014

Reducing computation in an i-vector speaker recognition system using a tree-structured universal background model

The majority of state-of-the-art speaker recognition systems (SR) utilize speaker models that are derived from an adapted universal background model (UBM) in the form of a Gaussian mixture model (GMM). This is true for GMM supervector systems, joint factor analysis systems, and most recently i-vector systems. In all of the identified systems, the posterior probabilities and sufficient statistics calculations represent a computational bottleneck in both enrollment and testing. We propose a multi-layered hash system, employing a tree-structured GMM–UBM which uses Runnalls’ Gaussian mixture reduction technique, in order to reduce the number of these calculations. Moreover, with this tree-structured hash, we can trade-off reduction in computation with a corresponding degradation of equal error rate (EER). As an example, we also reduce this computation by a factor of 15× while incurring less than 10% relative degradation of EER (or 0.3% absolute EER) when evaluated with NIST 2010 speaker recognition evaluation (SRE) telephone data.

42 ENGINEERING↗