Abstract
The performances of digital electromyographic signal processors in dynamic conditions are determined by evaluating the root-mean-square value of the estimation error, including bias and variance of the estimates. The results thus obtained are valid for a wide class of myoprocessor schemes. The analysis has been used to optimize myoprocessors with moving-average smoothing filters, the length of which will be adapted to the local variability characteristics of the force in order to obtain a minimum overali estimation error.