Abstract: A system is presented for generation of output sounds having psychoacoustic qualities comparable to input sound or sounds. Short term and intermediate term features are computed for each input sound, sound components are clustered, filtered, and scored; and a prediction learning system is trained on the probabilities of classes of regions over time. A decoder can make use of this information to generate outputs that sound similar to, but not the same as, the input sound or sounds. The method and apparatus can be operated with no special training.

Abstract: In the particular embodiments described in the specification, the velocity of a flowing liquid is measured by emitting acoustic signals from an acoustic signal unit into the flowing liquid and detecting signals reflected from solid particles moving with the liquid and a Fast Fourier Transform is performed on the Doppler frequency signals to produce a power signal spectrum. The resolution of the power signal spectrum is expanded at the high frequency end by taking the square root of the power signals and a least squares curve fitter produces a trend line from the data points at the upper end of the spectrum. The trend line is extrapolated to the zero amplitude axis to produce a signal representing the maximum velocity of the flowing liquid and an algorithm based on the shape of the conduit through which the liquid is flowing is used to produce a signal representing the average velocity of the liquid.

Abstract: In the embodiments described in the specification, an acoustic transducer unit emits acoustic signals into a moving liquid and receives reflected signals to produce Doppler frequency signals and a depth detector simultaneously detects the depth of the liquid. The Doppler frequency signals are processed in a Fast Fourier Transform unit to determine the average velocity of the liquid and a historical velocity versus depth table is prepared. Thereafter, each new velocity value is compared with an expected velocity value for the corresponding depth based on the historical table. The new velocity value is rejected if the variation from the expected value exceeds a selected variation. If accepted, the new velocity value and the expected velocity value from the table are weighted and combined to provide a new value for the table and also for a velocity versus depth profile unit. The velocity versus depth profile is adjusted based on the new velocity value using a least squares fit.