Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 5, 2014 16:57:25 GMT -6
If a plugin's algorithms benefit from oversampling, the developer takes care of it.
(Or at least, he *should* do so....)
In the real world, marketing says "more is better". To make you believe in this, they have to prove it. So they design converter chips, that sound better at higher samplerates. If a converter supports more than one samplerate, it will be optimized for one samplerate. Guess, which samplerate it will be......
|
|
|
|
Post by Bob Olhsson on Sept 5, 2014 19:02:22 GMT -6
Multiple sample rate conversions up and down often sound worse than just leaving the audio at a high rate until the very end. All acquisition is done at high rates. It's a question of when and how often it gets downsampled.
|
|
|
|
Post by jimwilliams on Sept 6, 2014 10:32:41 GMT -6
Dan says that you can't design a converter having better performance using real world parts than one operating at 60k. due to speed and precision limitations when you apply oversampling. This doesn't mean there aren't real world converters that perform better at 192 although it suggests they may not perform as well at 96k. as they ought to. That was several years ago. Since then many high test converters have been developed and released by TI and AD. Error rates are very low as is THD+noise. Many of these devices are not made for audio but industrial applications needing greater than audio bandwidths. 60k hz sample rate was selected because the converters available back then did degrade at higher than 96k sample rates. THD+noise increased. Now days those errors are not found in modern devices. What doesn't change is the larger the sampling bandwidth, the greater the noise floor and THD associated with it. This is why a converter's specs will degrade with larger bandwidths. It is no longer a result of limited silicon performance but the accumulation of larger measuring bandwidths. |
|
|
|
Post by Bob Olhsson on Sept 6, 2014 11:08:13 GMT -6
Dan said that the silicon speed performance of filters creates a theoretical limit on precision and he assumed oversampling was part of any overall system. His background is medical and industrial instrumentation. I agree that chips have thankfully come a long long way but I would never write off what he wrote to the noise and distortion limitations of older chips because that wasn't what he was talking about.
|
|
