Damn, I promised to post a productive algorithm

Hello,GriV but nothing myself In general, there is such a problem here. After encoding, for some reason I get a very dirty signal (for the encoder with adjustable compression) - no matter how I tried to get rid of the "dirt" - nothing it turns out. Therefore, at this point in time, the best encoder remains given above - only 9 seconds but the most excellent sound. I don’t have time yet, but I must put together a program for more productive (which is 1.5 times faster) encoder. And by the way, whoever will use it, I’ll give you advice on how to improve the results conversion. The fact is that the encoder uses a linear approximation, which generally speaking gives a very low quality of interpolation, and on average the encoder makes four samples per 1 word of the WAV stream. So open it a normal sound editor - for example soundforge - and interpolate the original one 44kHz 16bit stereo signal (most likely you will have one) at 150kHz [exactly this frequency - if exceeded, encoder errors are possible] 16 bit stereo and only after that, combine the channels (150 kHz 16 bit mono), and the resulting file slip it into the encoder program. I have compared the sound more than once, but there is one nuance - the program compresses the interpolated signal with a larger error (there there is an assessment of the compression quality) than non-interpolated, although it sounds the other way around - The sound quality simply improves significantly. Error calculation methodchanged, nothing changed, in short, the assessment can only be used in this way - than The smaller it is for a specific file taken, the better (-% approximation error In general, the parameter is insignificant - so you can safely throw it out (-%.