Hello Rob,that sounds interesting. But I don't quite understand the need for decimation.
Why go to the extremes of 32/or even 64 bits and 768 khz sampling and then decimate it to 44.1??
Any decent laptop can quite easily handle 32/768 these days.
Why on earth would one want to risk "throwing the baby out with the bathwater" by decimating a real HIRES format?
In today's world of ever increasing data capacities at lowering costs I just don't see any need whatsoever to decimate a theoretically as perfect and not bandwidth limited signal as can be captured. Why risk throwing away crucially important timing and possibly also spectral hf information by doing so.
I am aware that there can be problems with extreme out of band noise and digital artifacts aliasing down into the audible band. But it also a fact in the real analogue world of live acoustic music that both strings/brass and and especially some percussive instruments actually produce harmonics high above the rbcd 22khz bandwidth limit. With some percussion I think the harmonics reach into the 90-100khz region or even higher?
Please don't read this as a personal attack or criticism.
It is not in any way intended as such.
I am only asking these questions again because I really care about the best possible sound with digital recording techniques.
PS I hope you enjoyed your Machu Picchu trek.
Please educate me.
Cheers Christer
PS 2 You may run the risk of running into me again in London in July, I see Canjam there coincides with the PROMS.
I LOVE the PROMS and I am a bit curious to see what you have up your sleeve to be introduced at Canjam.
There are a number of reasons why a decimation
option is essential:
1. The demand for 44.1/48k is very strong - and there is a strong need for smaller bit rates and file sizes.
2. The Davina project is about research aas much as developing a product. One major question is the M scaler perfect? Or rather, how much of an audible loss is involved in bandwidth limiting>decimation to 44.1/48> M scaler back up to 705/768k. So the decimation filters are essential to answer this question.
3. Measurements of the ADC - the AP only accepts 192k as an input...
My personal view is that bandwidth limiting to 22 kHz is not a problem at all - the issues about high res is the timing of transients, and that can be solved with an appropriate tap length WTA filter. Of course, nobody knows what appropriate length is - that is at what point a doubling of tap length give no audible benefits.
Note the intent is to record at 705/768 then post process later to a required sample rate - so the baby won't be thrown out...
Yes I loved Machu Picchu, but the altitude sickness on the 4 day trek was not fun... I have been asked to post some photos, and hopefully will get round to it on Watts up? my blog.
See you at CanJam London!
I would have hoped Davina was higher priority, makes me sad to think of all recordings of great music made without better ADC technology. Also consider getting Davina ADC from tape of the old masters, finally perfect sound forever.
Yes I had already given myself the proverbial kick, so now most of my projects are complete, I will be putting much more focus on it.
I have the capacity to imagine all kinds of awesome things (probably best if I skip on the details) which generally only results in disappointment. There may be a silver lining however, as Rob’s Davina research might lead to tangible benefits in other products.
This is indeed the hope, and I would be extremely surprised if knowledge is not increased immensely- and more knowledge always improves new designs. What it will provide is certainty about the whole decimation/interpolation process. And certainty is a very rare commodity in the audio design game!