Abstract: Aspects of the present disclosure provide a method for training a predictor that predicts performance of a plurality of machine learning (ML) models on platforms. For example, the method can include converting each of the ML models into a plurality of instructions or the instructions and a plurality of intermediate representations (IRs). The method can also include simulating execution of the instructions corresponding to each of the ML models on a platform and generating instruction performance reports. Each of the instruction performance reports can be associated with performance of the instructions corresponding to one of the ML models that are executed on the platform. The method can also include training the predictor with the instructions or the IRs as learning features and the instruction performance reports as learning labels, compiling the predictor into a library file, and storing the library file in a storage device.

Abstract: The present invention discloses a voice signal synthesizing method and device, wherein voice signals are sampled at a relatively lower sampling frequency. During the reproduction of the signals, interpolation is used to calculate values of voice signals between two sampled periods and the calculated values are filled between the two sampled periods, whereby lower distortion rate may be obtained in the reproduced voice. This invention provides a low distortion rate and low sampling frequency voice signal synthesizing method and device.

Abstract: A simplified decoder for a decoder for 2-dimensional input devices with less number of pads is disclosed. Two pads are used to represent the addresses in one dimension of the input means in the 2-dimensional input device. Two capacitors are charged through a number of resistors, which number is decided by the number of addresses of said dimension. The time difference in charging the capacitors is measured to determine the address of an actuated input means in said dimension.