quote:

Originally posted by coroknight If you sample a sine wave at a low enough frequency it will turn into a triangle wave because of the lack of resolution. This effect becomes increasingly bothersome as the frequency goes up so it's not like super high frequencies are the only ones being effected.

I know. But this goes back to the same question I had before. How much will the 44.1 kHz standard actually distort the frequencies at, say, 13 kHz? Or 10 kHz?

Ive never heard any problems on a normal CD so I dont understand why anyone would need to render at any more than that. samplerate that is. bitrate though is a diferent case. 24bit actualy sounds better there. But it wont matter as the standard will be 16bit anyway. Its like trying to change the 16:9 format (only that that is easier).

Well after reading this thread, I decided to mess around. Obviously most of us know that you can't increase the perceived quality of a sample recorded at 44.1khz going to 96khz. But, if I go in my DAW and I have it at 44.1khz, open Gladiator synth, play a note, then, switch it to 96khz and play a note, I can hear a pretty decent difference, and I liked it better. But, funny thing, I don't use Nexus that often, but, I loaded Nexus, and guess what, it sounded worse when I went up in sample rate. Whats the deal?

___________________
www.soundcloud.com/timestretcher

quote:

Originally posted by cronodevir 2048 latency.

only way to have the best sound quality

___________________


https://soundcloud.com/ghostea stuff

quote:

Originally posted by ponsshin The drawing on top represents the 20 kHz signal being sampled at twice its frequency. 40 kHz seems fairly high enough, close to the frequency used in all mp3s but when you look at it, the 20 kHz is very poorly represented (only a couple values per period). The drawing at the bottom shows the same signal except it's being sampled at a frequency that's ten times as high. It seems crazy to sample something at 200 kHz but look at the way it's well defined now: 10 values instead of 2. Hope that helped

Ugh, this is exactly what people mean when they say "a little knowledge is a dangerous thing".

Please come back after you've read about sampling theory and interpolation algorithms. It would take far too long to explain it here, most people have to take a couple of years of signal processing before they really grok it.

___________________
My party schedule:
2009-02-21 - DJ Attention @ I'm So Popular
2009-06-18 - DJ Annoying @ People Need To Know Where I'll Be
2012-11-32 - DJ Insufferable ɸ Or At Least the Stalkers I Complain About
2048-06-66 - Spastic & Whocares ¶ Although I'm Actually Flattered
9999-45-81 - Tweaker Gimp ☼ I Probably Won't Even Go To This But I Have To Make Sure I Fill Up All The Available Space Here

It's up to each person to decide - some people will work at the samplerate the finished project will be rendered to - whether that's 44.1kHz for CD, 48kHz for DVD or 96kHz for blu-ray (not sure if that's the standard for blu-ray, I'm just going on a quick google search there). It seems that working at a higher bitrate has advantages when converting down to the final format, but working at a higher samplerate makes no difference when converting down.

If you're going to work at 192kHz, it's only going to make a difference if you release it at that samplerate.

On the other hand, if you have a computer that can handle it, and masses of storage space, there's nothing to say you can't work at 192kHz, 32 bit. If you're currently one of the world's best producers, there's a chance people may want the high resolution version in the future, and this will save you valuable minutes at that point, re-exporting the song at the higher samplerate (if you're using all software - otherwise it would take longer to re-record the external gear).

As for me, there are still many things I need to learn and improve before I start worrying about the small "perceived" difference that working at higher samplerates will provide me.

quote:

Originally posted by DigiNut Ugh, this is exactly what people mean when they say "a little knowledge is a dangerous thing". Please come back after you've read about sampling theory and interpolation algorithms. It would take far too long to explain it here, most people have to take a couple of years of signal processing before they really grok it.

Thanks for coming in Digi. People often compare audio to visuals, because they can get their heads around visuals - see, more pixels, better resolution, better quality! Same with sound - more kHz, better resolution, better quality! Um, no. As you say, people should read up on this stuff rather than going for a simple (wrong) explanation.

quote:

Originally posted by derail Thanks for coming in Digi. People often compare audio to visuals, because they can get their heads around visuals - see, more pixels, better resolution, better quality! Same with sound - more kHz, better resolution, better quality! Um, no. As you say, people should read up on this stuff rather than going for a simple (wrong) explanation.

There actually is an analogy to graphics if you think in terms of vectors as opposed to bitmaps. There's obviously a certain minimum resolution you need in order to display the correct angles, but once you go above that, you're not gaining any additional "clarity".

Although even comparing to vectors is an oversimplification. Start looking into heavy-duty image compression like NURBS and Bézier surfaces and such, and that'll give you a taste for how it works with audio.

You only need a very small number of data points to interpolate an audio signal, because sound is fundamentally sinusoidal and it's just a matter of figuring out which frequencies are present. Interpolation is generally not a causal system, meaning you don't need to get the answer in one cycle, you can get it in 10 cycles or 100 cycles, which is not even a millisecond of delay to the listener.

Edit: Forget the part about non-causality, I'm tired. They're easiest to model as a non-causal system but in the real world they obviously have to be converted to causal systems with delays. That's what I was talking about with the delays in the paragraph above.

___________________
My party schedule:
2009-02-21 - DJ Attention @ I'm So Popular
2009-06-18 - DJ Annoying @ People Need To Know Where I'll Be
2012-11-32 - DJ Insufferable ɸ Or At Least the Stalkers I Complain About
2048-06-66 - Spastic & Whocares ¶ Although I'm Actually Flattered
9999-45-81 - Tweaker Gimp ☼ I Probably Won't Even Go To This But I Have To Make Sure I Fill Up All The Available Space Here

Last edited by DigiNut on Jun-09-2009 at 00:20

I always think it's misleading when people compare picture resolution to audio sample rate. It's basic enough for people to get their head around but actually quite true for the reasons that Digi laid out.

If anything the example of the amount of pixels (as argued earlier) increasing resolution is far more analogous to BIT DEPTH, in terms of that example, rather than sample rate.

Again, I hate the analogy because it's far more complex and directly uncomparable, but think of when you see that slow motion video capture used for high speed sports such as formula 1 or tennis. When you see normal TV (30 fps) it looks fine, but when you the hi speed in real time (500 fps) it looks so much crisper.

So where resolution on a vertical plain is the amount of pixels or in this case as the comparing example, bit depth, then frames per second would be sample rate or resolution on a horizontal plain.

Now the argument (at least technically) for recording at higher sample rates is that if all it costs you is processing power and disk space then it could be worth the very minor increase in perceived quality.

Now I hate terms like this but in listening to accurately represented higher sample rate recordings such as 96/192k (in a pro environment), there is a smoothness present to the playback, which I attribute to greater interpolation (at least that's what my brain is telling me the reason).

Also, there are apparently sounds in things like high strings that present at higher frequencies (which would not normally be accurately captured at lower sample rates) but their lower DO harmonics present at lower frequencies, which are within our range of hearing. So when these are combined we hear a perceivable difference.

the main reason that we won;t need to work at higher foramts is that the 16b 44.1k is good enough for 99.9% of the population and was made that way for the reasons already discussed. If consumers want more (hifi enthusiasts etc), then they can go all analogue which is better than any digital system on the horizon anyway.

As for the OT, I don't know anyone that WRITES in a higher samplerate, but I do know some that record/track/overdub/sound design/mix at higher formats.

quote:

Originally posted by DJ RANN Again, I hate the analogy because it's far more complex and directly uncomparable, but think of when you see that slow motion video capture used for high speed sports such as formula 1 or tennis. When you see normal TV (30 fps) it looks fine, but when you the hi speed in real time (500 fps) it looks so much crisper.

No, no, no, no, no. In the case of motion picture it's your brain that's doing the interpolation, and your brain is a very rough and inefficient instrument for this. With audio it's the hardware doing the interpolation and the interpolation is essentially perfect.

If you want to compare to animation, compare it to the screen's refresh rate and not the frame rate. Humans have the ability to see flicker up to about 50-60 Hz; once you take it up to 75 or 90, it's not flickering anymore. It doesn't matter if you take the refresh up to 200 or 500 Hz, it will still look exactly the same, because that's what we can see.

But that's STILL a false analogy because you can film a 75 Hz screen with a 75 Hz camera and still see flicker because they're out of sync. On an audio signal it does not matter where you take the samples. As I said, the closest graphics analogy would be something like a NURBS curve, which can (perfectly) model a theoretically infinite surface with just a handful of control points and knot vectors. With audio, you only need the "control points" because you already know the "vectors" (sinusoids).

Please, no more misinformation on this. You can tell yourself whatever you want, but any differences you're hearing working in 192 kHz are more likely than not due to aliasing distortion on the down-conversion and weird compatibility artifacts.

___________________
My party schedule:
2009-02-21 - DJ Attention @ I'm So Popular
2009-06-18 - DJ Annoying @ People Need To Know Where I'll Be
2012-11-32 - DJ Insufferable ɸ Or At Least the Stalkers I Complain About
2048-06-66 - Spastic & Whocares ¶ Although I'm Actually Flattered
9999-45-81 - Tweaker Gimp ☼ I Probably Won't Even Go To This But I Have To Make Sure I Fill Up All The Available Space Here

Digi, why is it then, if I run Nexus it sounds worse, which goes with what your saying. But, when I run Gladiator at a higher rate, you can hear that it sounds better? Its more crisp and defined or deeper sounding. What I am talking about is not dithered down, which is one thing that I dont know if your refferring to with the aliasing. I thought u meant working in a higher rate, not sure. But, Nexus working in a higher rate does sound like it actually cuts away frequencies.

___________________
www.soundcloud.com/timestretcher

quote:

Originally posted by DigiNut No, no, no, no, no. In the case of motion picture it's your brain that's doing the interpolation, and your brain is a very rough and inefficient instrument for this. With audio it's the hardware doing the interpolation and the interpolation is essentially perfect. If you want to compare to animation, compare it to the screen's refresh rate and not the frame rate. Humans have the ability to see flicker up to about 50-60 Hz; once you take it up to 75 or 90, it's not flickering anymore. It doesn't matter if you take the refresh up to 200 or 500 Hz, it will still look exactly the same, because that's what we can see. But that's STILL a false analogy because you can film a 75 Hz screen with a 75 Hz camera and still see flicker because they're out of sync. On an audio signal it does not matter where you take the samples. As I said, the closest graphics analogy would be something like a NURBS curve, which can (perfectly) model a theoretically infinite surface with just a handful of control points and knot vectors. With audio, you only need the "control points" because you already know the "vectors" (sinusoids). Please, no more misinformation on this. You can tell yourself whatever you want, but any differences you're hearing working in 192 kHz are more likely than not due to aliasing distortion on the down-conversion and weird compatibility artifacts.

er....well that was my whole point. Which bit of this didn't you get:

quote:

Originally posted by DJ RANN I always think it's misleading when people compare picture resolution to audio sample rate. It's basic enough for people to get their head around but actually quite true for the reasons that Digi laid out.

I just made the point that if you HAVE to use some sort of analogy for the layperson, then pixel count is more comparable bit depth and FPS is more comparable to sample rate.

I don't think enough laypersons would understand the difference between refresh rate and FPS, whereas people with a very basic or surface understanding how TV's (or flickbooks) and audio works will get this analogy. That's all

technically you're analogy is "closer" but as I said before you simply can't compare the two.