Higher sample rates in Pro version

Very funny post. However, my question is not that stupid as you suggest, and I would just like to have some input from people that actually listened to the higher sample rates.

Cheers,

Jack

I'm not ridiculing nobody.  Please don't take me wrong. 

To notice any suposed difference, suposing someone had a hearing up to that, would require really very good sound monitors, and high quality soundcards.

I noticed a difference when recording my digital piano at 44/16 vs 88/24. At 44/16 there were "holes" in the sound. Hard to describe, but there was a certain lo-fi quality to it. Though I don't know whether it was the 88kHz sample rate or the 24-bit depth or both that was responsible for the improved sound.

But with Pianoteq, the sound is generated in the digital domain. There's no analog-to-digital conversion. And 44kHz/16-bit sounds fine. I think 96kHz is the standard for films though. But for the time being, 48kHz can easily be converted to 96kHz.

You are going to be a happy man, mooks,
because we can give you the scientific explanation for why 96k could possibly sound 'better' (? different to 48k right now!

it is this:

Doubling the frequency up to 96k produces an extra octave of supersonic sounds. Although your ears cannot detect these sounds directly,
these soundwaves can and do interact with each other and the rest of the piano sound (e.g. sum and difference tones), and can thus in principle produce audible partials within the frequency band that your ear *can* detect.

So there you go

The most dramatic example of this principle is the electronic instrument called a "theremin", I'm sure you've heard one?  IIRC, this produces two very high frequency tones, however when your hand moves about and alters one slightly due to changing the capacitance, you can hear the difference tone move into the audible band and change in pitch

Yeah, OK, I've heard something along those lines before! I just don't buy it. I want some evidence. Maybe someone could provide me with some supersonic samples that I could mix and hear some wonderous audio. Or perhaps we could pipe 96kHz Pianoteq through a supersonic high pass filter and see if we can still hear anything.

If I understand right, the generation of lower frequencies must be a non-linear effect. So it must be happening in the amp and speakers, and the ears if any high frequencies survive. If there is an effect, it's probably highly dependent on your equipment and you could be hearing something quite artificial. If there is an ideal result, then maybe that could be encoded in a plug-in that takes the 96kHz and spits out a richer 48kHz!

Anyway... maybe I should stop worrying and go and write some music...

You don't believe in mathematics and the principle of linear superposition of waves?!?

How do you think Pianoteq works? Do you think there are little men inside your computer, playing a tiny invisible piano?

With nano-technology, maybe one day!

I certainly do believe in mathematics. What I was getting at is that the "interaction" explanation explains why we *might* hear some extra stuff at 96kHz, but I just don't yet accept that we actually do. My suspicion is that those effects are insignificant. Hence why I'd like to see some evidence.

I tried an experiment. I generated two sine waves of frequencies 22kHz and 23kHz at 96kHz sample rate. I played them through a soundcard and amplifier that can both apparently go up to 40kHz. I turned up the volume, but couldn't hear a thing over the background hiss.

Yeah, I can't do that experiment either, but what we can do is compare 48 and 24 in the same way, and playing them at double speed might indicate what you'd hear for a 48/96 comparison, although sounding like a toy piano. Anyway, I tried it and the two don't really cancel out much at all. I don't even get the same result when I record the same sample rate twice (using the same midi track).

Which brings us back to my original question: are there any Pro users out there that hear a noticeable difference at higher sample rates?

As to the discussion whether or not we can hear supersonic detail, I think possible differences in audio quality at different sample rates has more to do with the roll-of filters in AD/DA converters. These are less then perfect, and at 44.1 kHz may introduce audio artifacts in the audible spectrum. By using higher sample rates, if I understand this correctly, the audio artifacts are shifted to frequencies that only dogs can hear (there you go, Beto-music). 192 kHz is definitely overkill, but 88.2 or 96 kHz can make a noticeable difference.

Cheers,

Jack

P.S. I am monitoring through Genelec 8040's using Prism Orpheus converters.

You're right, Mark, we shouldn't worry about this when it comes to using Pianoteq. I think Pianoteq sounds fantastic and if it improves, it will be due to Modartt improving the model, not because of sample rate.

But I still want to satisfy my intellectual curiosity. I have a background in physics, and have always had a technical interest in audio. I need to know what's going on!

Anyway, you promted me to do some more research, and as Funkasizer says, it all seems to be about signal processing, either in the digital chain, or the DAC roll-off filter. CD players etc will typically over-sample by a factor of 8 to get round these problems. Also, any processing you add in the digital signal path (EQ, effects, reverb etc) probably has less than perfect implementations of digital filters, and running the whole system at a higher sample rate can reduce the errors that they introduce (but be warned, some processors can apparently add more noise at higher sample rates!). At the end of the chain, dithering to 16bit/44.1kHz should be fine for actual listening, unless you have dog ears.

It seems that the same benefits could be obtained by running a 48kHz Pianoteq through a high quality 96kHz upsampler before it passes through the rest of the chain. Maybe that would use less CPU, but if you *can* afford to run the source at 96, then why not.