65.19.43.115
In Reply to: RE: Bad idea unless you can hear like a bat. posted by Paul Folbrecht on June 07, 2009 at 14:32:38
Just because there are plenty of 'knowledgeable people' that disagree with me, I must be wrong. That’s the story of my life.
First of all, it is critical to understand that I was talking about the RECORDING sample rate, not PLAYBACK. The criteria for each in regards to audible fidelity are very different. There is no doubt 44.1K/16 is insufficient for both recording and playback. It is my contention that 88.2K/24 is sufficient for RECORDING LPs. Other than ticks and pops, there is no supersonic content of any consequence on an LP. The typical phono cartridge has enough trouble tracking 18KHz let alone 20KHz and beyond.
Nyquist theory deals with ALL waveforms. It is the practical implementation of the theory that screws things up. At every step, sampling theory demands band limited signals but, in practice, perfect band limiting it very difficult to achieve. In theory, 44.1K is sufficient to record all audio up to 20KHz but the filters necessary for the prefect band-limiting required by the theory have audible side effects. In the lab, with steady-state sine waves, 44.1K measured perfectly but not so with music due to the phase shifts and other anomalies introduced by the brick-wall analog filters.
That realization led to resampling and digital filters. With a digital filter, each output sample point is computed from a weighted average of the nearby input samples. The weighting coefficients determine the characteristics of the filter. In theory, digital filters are perfect and, in the lab with steady-state sine wave, they measure perfectly, as well. But music is not steady state and the averaging effect of digital filters takes its toll, such as with transients and every time there is a change in frequency.
The current state of digital audio means choosing from a spectrum of audible poisons. Audiophiles with a low tolerance for digital filter poison choose NOS and no filters. Others with a low tolerance for NOS poison choose resampling and digital filters. The reason many prefer hi-res recordings is they get the benefits of high sample-rate playback without the poison of resampling and digital filters. The cost is a huge dataset. Pick your poison.
My ideal is to start with lowest practical recording/storage bandwidth, 88.2K/24 for LPs and 44.1K/16 for CDs, and resample to a higher rate for playback without using digital filters. I have been working on this for many years and I now think I have the technology to do it. The first step will be to resample CD files to 88.2K/24 on the PC and feed both LP and resampled CD files to a custom DAC that will resample an additional 1x to 32x using both digital and analog techniques. The second step will be to resample both LP and CD files to 176.4K/24 on the PC and feed the same DAC. I suspect the poisonous effects of digital filters are most noticeable at lower sample rates: That’s why I think I can get away with using traditional digital filter to go from 176.4K to 705.6K and beyond. We shall see.
Follow Ups:
Post a Followup:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2009, Audio Asylum, all rights reserved ]
[ Go Green ]
[ Take5 ]