Hi Pat, > The real trick is dithering the data back down to 16 bits. > Just truncating the data (either double, single floating, > or long integers) is clearly audible.
Hmmmm...clearly audible? Can you elaborate? According to Craig Maier at Diamond Cut the worst you'd see for 16 bit red book would be 0.0008% THD. That's if the program got it wrong 100% of the time. Full quote... > Basically, talking about rounding errors in the same breath as using the > word Vinyl is humerous to me. Relatively speaking, there is no impact > unless you transferred the recording at around - 50 dB or lower into a > 16 bit system. Lets consider the performance of the various components > involved in the process of record transfer: > > 1. Vinyl LP Record: 60 dB dynamic range at best > > 2. 16 bit Red Book CD Audio dynamic range: 96 dB > > 3. Diamond Cuts Math routines dynamic Range: 384 dB > > The rounding error will be +/- 1/2 LSB (Least Significant Bit) in > general. That translates into the following: > > 1. Vinyl LP: 1 part in 2000 (assuming that it was a digital system - - > - there is no real rounding error in an analog system like a vinyl > record, but any signal below the noise floor is of no value - - - thus > anything below 60 dB is useless. But I show its dynamic range in ratio > form for your comparison to digital systems shown below) > > 2. 16 bit Red Book Audio: 1 part in 64,000 which represents about > 0.0008% Harmonic Distortion. The best Vinyl records had 1 to 2 % THD. > > 3. Diamond Cuts Math routine: 1 part in 3.7 x 10 ^19 > > So, you can see that the weak link in the system is, by far, the vinyl > LP record. You will never in a million years hear this quantization > error unless you transfer the record down in the mud of the system (VU > meters not budging at all). > > Also, I will pay a kings ransom to anyone who can show me that they can > discern a difference due to quantization errors introduced by gain > normalization, but they must use double blind testing / listening > techniques to prove their point. > > It is just math and science; there is nothing magical going on here and > common sense should dictate a reasonable conclusion. http://www.diamondcut.com/vforum/showthread.php?t=1845 It's humorous to me that the fella that started this thread sounded like he'd just about be satisfied recording his records through a VOIP phone line, and here we are splitting the hair to the nth degree. :) Regards, eCo -- eCo ------------------------------------------------------------------------ eCo's Profile: http://forums.slimdevices.com/member.php?userid=4408 View this thread: http://forums.slimdevices.com/showthread.php?t=21836 _______________________________________________ audiophiles mailing list [email protected] http://lists.slimdevices.com/lists/listinfo/audiophiles
