View Single Post
Old 18 October 2016, 23:08   #118
Registered User
Join Date: Jun 2010
Location: PL
Posts: 1,653
Originally Posted by meynaf View Post
So you never write about poor Amiga audio quality. But you still wrote that it just can't do 14 bit, that it's closer to 10. You say there is noise, which not only me but several other ppl here don't hear (even when formatting floppies, lol). And you didn't disagree with anyone saying it was crap (my bold quotes).
However if you're vaccinated then why not just stopping here and try to discuss something more interesting.
I never said crap also i can't take responsibility for things said by others but i can understand sometimes their reaction - from time to time it is very difficult to not realize that you behave in a condescending way - some people not dealing with this gently.
From my perspective this very amusing part of science applicable as in theory you can imagine such software approach (you should like this - there is no hardware but pure software) to beat limitations of hardware - for example deliver audio quality beyond Amiga designers imagination.
List thing more interesting than breaking old HW limits - i think this is important especially for you as a software guy - i have own ideas about improving HW but it is real art to not change main HW architecture yet bend limit beyond imagination only by using clever software.

Originally Posted by meynaf View Post
Of course ears differ about perception of small details. But big noise is big noise for everyone, isn't it ?
If you want to experiment oversampled 8 bit with dithering and aggressive noiseshaping, you can tell me exactly what computations have to be made on the signal, and i can try coding it. I have the code (somewhere) to play audio by cpu poking AUDxDAT.
Yes and no - it depends from many things, side to this our auditory system can adapt and begin to ignore some distortions (noise like mild hiss can be one of them).
Computation you can check for example SoX or ffmpeg code (ffmpeg use SoX part but they added coefficients for some common sub-Nyquist sample rates - disputable if they improve or not perceived quality but they exist).
Someone provided code for using Copper (as my assumption is that Copper will provide jitter free sampling) but even using HW DMA and sample rate like 56kHz with proper signal processing i think Amiga can easily match dynamic of the 16 bit system (with respect to Gerzon-Craven theorem).

Originally Posted by meynaf View Post
Ok so you want to measure sound quality. But, perfect accuracy in comparison to the digital signal, wouldn't lead to quality. The replay frequency itself (and preferably its harmonics as well) must be removed from the sound (aka anti-aliasing), so it won't be 'accurate'. I doubt any output in the world do exactly what the numbers say : if the samples say f.e. 1,3,5, values 2 and 4 should and will be observed in the output.
So with all this, what do you want to do ?
Nope - anti-aliasing is a problem of AD conversion and later can't be removed from signal - DAC doesn't require anti-aliasing filter - DAC require reconstruction filter (sometimes it must be also anti-imaging filter).
There are some techniques to deal with problems, temporal averaging to remove uncorrelated with signal noise is one of them. And also we note measuring particular sample values (albeit it possible and it is used with some DAC calibration techniques).

Originally Posted by meynaf View Post
Didn't i read that already ? Oh well, i'll take the time to check that again.
Obviously not as you arguing using arguments common for some audiophiles - once again i never heard on equipment that has bad objective (measurements) results but anyway has amazing sound - behind good sound reproduction there is lot of scientific knowledge.
Wise people not say f**k theory - i live long enough to experience situation when good theory perfectly match real life.

Originally Posted by meynaf View Post
Well, looking again, it seems my PC has no RCA input. And even though it can perhaps outperform 14 bit, it's still just 16. Input A/D may filter or otherwise damage the signal. Rather than a PC, an oscilloscope may give better results - with samples tuned for that use - but i am not equiped.
Nope, practically all audio PC cards using delta sigma AD converters - they are free from aliasing problem also AA filter is not a limitation as due way how DS works those filter are quite gentle and they are stable in phase domain.
Oscilloscope will be highly unsuitable for this as very limited dynamics present in typical oscilloscope - they usually use not more than 8 bit AD converters and with typical sample rates around 100MHz efficient number of bits is usually somewhere around 6 or less - definitely very bad for 14 bit audio.

Originally Posted by meynaf View Post
I have written a program that can play wav, aiff, and flac (and mp3, and even ogg if i enable them). Custom versions can be made for different output comparisons. However you won't be able to use it on A500 and i'm not sure tests made on an emulator can be meaningful.
Nope - emulator is not interesting as we testing idealistic not real system that functionally is close to Amiga but completely different.

I have A1200 and A3000 (CPU capable to run 68020 code) - in 3 -4 weeks i should e able to use A1200 (need to buy IDE>SD adapter and restore system disk), A3000 is slightly more challenging as i need to rework some problems on MB and buy a keyboard - as you can imagine it is impossible to buy A3000 keyboard nowadays.

Originally Posted by meynaf View Post
I'm not saying every part of circuitry can't affect audio signal. I know they can.
But having plugged it into some big amp where normal music with volume knob set to 8 would have been enough to make the neighbors come, i set it to 40 and still got only silence. Let's be honest : i was very surprised to hear nothing !
We not talking about high fancy amp - i know what I've heard on some Amiga's mentioned A600 is disaster not only due leaking caps.

Originally Posted by meynaf View Post
Maybe you're right and it's not good by todays standard. Yet it sounds good in everyday life.
If you play 24-bit samples on super accurate 24-bit D/A, will you hear the difference with regular 16-bit ? Where's the point in which it starts to count ?
If 14-bit doesn't sound different than 12-bit then maybe we only have 12-bit dynamics while playing 14. But then whether we have 12,14,16 or more, doesn't matter (at least for me).
But this was never my goal to match Amiga with modern digital audio - it will be unfair as Amiga is 8 bit, it is designed in particular way and it was compromise between cost and other aspect of computer - this is same as watching movie at Atari ST or Amiga - it is not about cinema experience more it is how much of things we can squeeze from those machines having present knowledge - i remember when played with DSP and samples on Amiga and i know how moronic was almost everything from this time yet thanks to this times i evolved to current state.
And from beginning quite clearly stating that my listening experience telling me that Amiga is somewhere between 60 and 70dB which is not bad and i will be very positively surprised to see better numbers.

Originally Posted by meynaf View Post
Weren't you the one with the proposal to stop talking ?

Anyway, we have to remember that we communicate in english, which i don't regard as a very good tool, especially when neither is a native speaker.
So many misunderstandings can occur.
Yes but then side discussion was triggered about analog circuitry around Paula - i've expressed my point of view on current hype that every company currently trying to create impression having fancy audiophile equipment only due of fact that you can replace OPAMP in socket.
And there is large group of people seriously deliberating on superiority of Analog Devices AD8597 over AD797 - referring to Amiga - there is no magic - simple OPAMP replacing will not magically change sound - many thing need to be addressed to improve sound - this is another interesting topic for Amiga - after checking what is better - 14 bit or 8 bit with oversampling and NS there will be time for some improvement on Amiga to even improve audio more.
And i agree on language (my English is really self learned - i know that my vocabulary and grammar are poor), i agree on stopping discussion to the point where some objective data will be known so i t will be easier to see some picture and discuss something more real than our impressions.
Once again - i respect you even if sometimes it is not easy to discuss with you.

Last edited by pandy71; 18 October 2016 at 23:18.
pandy71 is offline  
Page generated in 0.04119 seconds with 10 queries