Then I started counting

Hello TomCaT and search for relevant literature. It turned out like this, not a very pleasant thing - The last of the attached algorithms is the best and here's why. As my respected colleagues have already pointed out to me, DSP is characterized by such a thing as aliases that are a consequence of frequency maxing or frequency substitution (analogous to the strobe effect) - when the real sound different from what it should sound like. So, these same aliases in Depending on the playback frequency of the already converted sound, stronger/weaker influence on the final result, and the last sample presented The digitized signal is almost in the optimum zone. Speed increase reproduction (which I carried out, although the technique turned out to be bad) led to a deterioration in sound, although logic dictated otherwise - with increasing speed of work, the quality of what is reproduced should increase. That. you can only slightly improve the quality of the reproduced signal by slightly changing the operating code (by adding a couple of NOPs in the right places), you will qualitatively improve the real Speck sound is not allowed. To summarize, I conclude that for the basic (non-turbo) configuration there is almost ideal algorithm (((((((-;