Patent №
US 6,163,608
Granted
2000-12-19
Filed 1998
Owner
ERICSSON INC.
Lab
—
AI components
2
speech · planning
Assignment
Recorded
Dataset
AIPD
2023_r1 edition
Application
09005145
Methods and apparatus for parametrically modeling background noise and generating comfort noise in echo suppression systems are disclosed. According to exemplary embodiments, a background noise model is based on a set of noise model parameters which are in turn based on measurements of actual background noise in an echo suppression system. The exemplary embodiments include autoregressive, autoregressive moving-average and frequency-domain models. An exemplary first-order autoregressive moving-average model includes a single fixed zero and a single variable pole. The single zero and the single pole are sufficient to provide appropriate spectral tilt in the resulting modeled noise, and the single zero ensures that the model is unconditionally stable. Integration of the exemplary parametric noise models with various echo suppression devices is also disclosed.
AI classification
Ownership
ERICSSON INC.
assignment · 89500673
Assignors
ROMESBURG, ERIC DOUGLAS, BLOEBAUM, LELAND SCOTT, GURUPARAN, CORATTUR NATESAN SAMBANDAM
On an employer assignment, the assignors are typically the inventors.