Project DetailsWhen sampling an audio signal and converting it to a digital, discrete signal, we face several challenges that put the samples at risk of loss of information. One of those risks is driven by the dynamic range of the sampler. A signal that crosses the dynamic range is clipped, and therefore its reconstruction is damaged. In the following discussion, we are about to introduce analyzation results of Beyond Bandwidth Residual Reconstruction (22) algorithm implementation.
Here we propose a robust, real-time algorithm that is designed to sample signals that cross the dynamic range with minimum loss of information, relying on an input signal that had been activated by a modulo operator. We consider testing several aspects of this algorithm and contrast its performance to a classic sampler (where clipping is allowed) and the Automatic Gain Control () algorithm for pre-sampling detection, which is a well-known method for altering the input signals amplitude to adjust to the dynamic range. By empirical study, we show that 22 solution can dramatically increase performance compared to AGC detection followed by classical sampler. Furthermore, this study shows that separating the input data into small batches can improve the existing 22 algorithm by making it a real-time algorithm.
