Is anyone still working at the 44.1khz sample rate?

[EdGrip]

As mentioned above, I can't imagine SoundCloud are streaming things at 96kHz.

[bxbrkrz]

Good idea. First, I would record the sound of that vinyl from a Gramophone's horn, with a Milab VIP50.

I always feed my final audio through an old shoe too

I need those headphones right now

[guitfnky]

I'm gonna share this last comment here in this thread, then I'm out of the discussion, because honestly I hate this topic. Feel free to comment or disagree with me, but I won't reply. My mind is already made up about this.

I couldn't care less about the science, logic or explanations behind all this sample rate crap. Most people say sample rates doesn't make an audible difference. For me it does. Maybe it's the way the plugins process shit, or the way Reason handle things, or maybe my sound card does something weird. I don't know and I don't care. All I know is when I bounce or play my music in Reason at different sample rates, there can SOMETIMES be audible differences.

Here's a snippet of a track I'm working on.
For me the 96000 Hz version has a fuller guitar sound, like it's got more body compared to the 44100 version. (The "guitar" is a processed synth btw, so it's not a recorded sound).
I mean listen to the difference in the guitar tone at 0:22 between the 44100 and 96000 version. To me it sounds like it's mixed differently. And it's not that particularly subtle either. The 96000 guitar tone sounds more powerful imo.
The 44100 version has a brighter sound on hihats and cymbals. To me that's pretty clear. It's a got some sort of sizzle sound on the hihats that I don't like, like it's not processed in a natural way.
The 88200 doesn't have that annoying sizzle that 44100 has. 88200 Hihats and cymbals are just a tiny bit brighter compared to 96000 version. It almost feels like the 96000 has the hihats turned down slightly in the mix compared to the other two.

So anyway I know people will claim that it doesn't matter. I typically go for 96000 Hz myself, beacuse I usually find it less harsh sounding. But for this track I might actually prefer the 88200 version.

I know you won't do it, but I'd love for you to make these downloadable so we could do a null test.

[TritoneAddiction]

I know you won't do it, but I'd love for you to make these downloadable so we could do a null test.

They should be downloadable now. Have fun

I'll probably keep the examples up for a couple of days (if anyone wants to test them) and then I'll remove them.
Since it's not a finished track I'd rather not have them up on my channel for too long.

[guitfnky]

I know you won't do it, but I'd love for you to make these downloadable so we could do a null test.

They should be downloadable now. Have fun

I'll probably keep the examples up for a couple of days (if anyone wants to test them) and then I'll remove them.
Since it's not a finished track I'd rather not have them up on my channel for too long.

thanks! the audio files are definitely different--the timing isn't the same for one, and I think the level may differ a little too, so I can't really do a proper null test (I've learned that trying to manually line stuff up is a fool's errand). maybe one day I'll try doing an export of one of my own projects at a few different sample rates to see if I can get a good test.

[TritoneAddiction]

They should be downloadable now. Have fun

I'll probably keep the examples up for a couple of days (if anyone wants to test them) and then I'll remove them.
Since it's not a finished track I'd rather not have them up on my channel for too long.

thanks! the audio files are definitely different--the timing isn't the same for one, and I think the level may differ a little too, so I can't really do a proper null test (I've learned that trying to manually line stuff up is a fool's errand). maybe one day I'll try doing an export of one of my own projects at a few different sample rates to see if I can get a good test.

I've used the exact same file for all of them, only rendered it at different sample rates. So any possible volume change is the result of different sample rates. The timing should be the same. Again I haven't done anything differently to them.
Also I've worked with synths here, so there's gonna be different "takes", since the synths have lfos running and stuff like that.

Just use your ears man. The differences should be fairly obvious. Look for the things I wrote earlier, especially the hihats.

[guitfnky]

thanks! the audio files are definitely different--the timing isn't the same for one, and I think the level may differ a little too, so I can't really do a proper null test (I've learned that trying to manually line stuff up is a fool's errand). maybe one day I'll try doing an export of one of my own projects at a few different sample rates to see if I can get a good test.

I've used the exact same file for all of them, only rendered it at different sample rates. So any possible volume change is the result of different sample rates. The timing should be the same. Again I haven't done anything differently to them.
Also I've worked with synths here, so there's gonna be different "takes", since the synths have lfos running and stuff like that.

Just use your ears man. The differences should be fairly obvious. Look for the things I wrote earlier, especially the hihats.

wasn’t suggesting you’d manipulated them. I assumed there was something else that caused the differences (whether that’s how Reason exports at different sample rates, etc.). good point about the LFOs.

and the problem with just using my ears is that they’re inextricably linked to my brain. as accurate as any one person’s ears may be, they’re still attached to that gray blob between them, with all its own imperfections. notoriously untrustworthy, especially when it comes to perception—even more so when primed by instructions on what to look out for.

[TritoneAddiction]

I've used the exact same file for all of them, only rendered it at different sample rates. So any possible volume change is the result of different sample rates. The timing should be the same. Again I haven't done anything differently to them.
Also I've worked with synths here, so there's gonna be different "takes", since the synths have lfos running and stuff like that.

Just use your ears man. The differences should be fairly obvious. Look for the things I wrote earlier, especially the hihats.

wasn’t suggesting you’d manipulated them. I assumed there was something else that caused the differences (whether that’s how Reason exports at different sample rates, etc.). good point about the LFOs.

and the problem with just using my ears is that they’re inextricably linked to my brain. as accurate as any one person’s ears may be, they’re still attached to that gray blob between them, with all its own imperfections. notoriously untrustworthy, especially when it comes to perception—even more so when primed by instructions on what to look out for.

It's fine. I get it man. I understand your point. I really do. But sometimes you also got to trust your senses. I know they can play tricks on us, but sometimes they do tell the truth. If we're taking my examples here, for me there's no question that they sound different. I've experienced examples where I couldn't tell the difference. But this isn't one of them. Like you say, maybe it's the way Reason handles it. Who knows.

[integerpoet]

The question isn't whether there's a difference. There absolutely is and it's not even worth listening for; folks ought to just trust you when you say you hear it. But that's not the question; the question is whether the difference is meaningful, which might sound judgmental but isn't meant that way. Here's a concrete example of what I mean:

If your configuration produces unintended audible aliasing, you might be able to mask that problem with a higher sample rate, and you might say: "Good enough. It's easy and it works and I don't care about file size and I need to move on the next project." Which is totally fine because engineering is about trade-offs and time is money.

But that doesn't mean a higher sample rate inherently sounds better. It just means you didn't have time or patience or interest or whatever to track down what was producing the aliasing and correct it so you could export at a lower sample rate without hearing the problem.

Disclaimer: my audio engineering (making products as opposed to twiddling knobs) is many years out of date and dimly remembered and I at this point I am mostly going by this video. At least I think this is the right one. I didn't watch it all the way through again to be sure.

(Every time he says "keela hertz" I think of "gila monster" and snicker because I apparently have an emotional age of 12.)

[orthodox]

It just means you didn't have time or patience or interest or whatever to track down what was producing the aliasing and correct it so you could export at a lower sample rate without hearing the problem.

Agree 100%.

[Rising Night Wave]

may i say something?

higher hertz and bit depth means moer smooths signal when converting from digital to analog -> to speakers.

if i listen to a decent amount of time let's say all the time at 44,1/16.... let's say for a period of ... i don't know... 5 years?... then when i switched to true 192/24 when i bought external usb sound card... WOOOOOOOOOOW it sounds so different. more rich sound, more smooth sound, more detailed sound.

that is why i choosed that format from then on.

other thing is - solely because i read on internet on many places - higher hertz and bit depth make hard fxing the song/track more precise. less ugly sounding. more good results. that is why some DAWs offer even much higher sampole rates and bit depths. cakewalk. amazing!

conclusio: if you have a great speakers or hii-fi or even high-end system... o ho ho, then these higher rates brings song to life. power and so on...

ps: i also noticed, using reason... let's say in real time play...when using more sounds with higher sample rate ... there is more space for all those sounds. they become more distinguish in all that mixed end song. it's simple as that.

[chaosroyale]

Again, this is a very common misunderstanding, but it is a misunderstanding of sampling theory. Watch that video, seriously!
As I said earlier, as long as the sample rate meets the minimum requirements, a wave will be reconstructed perfectly. This is because there is one, and only one, mathematical solution for a given set of sample points for band-limited signal.

Anything else you hear is coming from your specific setup, such as poorly designed filters, VST plugins which need oversampling, or noisy cheap connectors on a sound interface. In your case, you changed your hardware! And maybe your distortion plugins alias less at the higher rate. If you like that sound, fine!

But for actual playback of digital audio, both 44.1 or 48 kHz perfectly reconstruct the full frequency range.

As for dynamic range; 16 bits is more than enough for pro-level playback at home, in the cinema, or live. Remember, all the classic albums recorded on reel-to-reel tape in the 70's and 80's have less than 16 bit dynamic range.
24 bits is waaaaaaay more than enough in all cases and gives you a ton of headroom for export if you are sending to a client who requires that.

may i say something?

higher hertz and bit depth means moer smooths signal when converting from digital to analog -> to speakers.

if i listen to a decent amount of time let's say all the time at 44,1/16.... let's say for a period of ... i don't know... 5 years?... then when i switched to true 192/24 when i bought external usb sound card... WOOOOOOOOOOW it sounds so different. more rich sound, more smooth sound, more detailed sound.

[Carpainter]

The 44100 version has a brighter sound on hihats and cymbals. To me that's pretty clear. It's a got some sort of sizzle sound on the hihats that I don't like, like it's not processed in a natural way.
The 88200 doesn't have that annoying sizzle that 44100 has. 88200 Hihats and cymbals are just a tiny bit brighter compared to 96000 version. It almost feels like the 96000 has the hihats turned down slightly in the mix compared to the other two.

Have you tried exporting at a higher sample rate and then reducing the sample rate to 44.1k in a different program? This will negate any aliasing that you might be getting at 44.1k in Reason, if that's where the problem lies.

One more thing to consider: Are any of your drum samples being downsampled or upsampled? If I set my sound card to handle downsampling and upsampling, I can sometimes hear a subtle difference on the high end (cymbals and hats).

[guitfnky]

The question isn't whether there's a difference. There absolutely is and it's not even worth listening for; folks ought to just trust you when you say you hear it.

not necessarily true. a difference while in the box can be fully negated when it comes back out. and it’s physically impossible for us to hear what’s going on inside the box. as others have already explained, audio waveforms are recreated perfectly up to half the Nyquist frequency. the only audible differences are introduced by other factors (anything randomly introduced each time a file is bounced, aliasing, and the like).

as for why people shouldn’t just take others at their word, that seems a silly argument, when there ways available to objectively prove one way or another whether there’s a difference. that’s to ask “why prove it when you can take it on faith?”

[visheshl]

The question isn't whether there's a difference. There absolutely is and it's not even worth listening for; folks ought to just trust you when you say you hear it.

not necessarily true. a difference while in the box can be fully negated when it comes back out. and it’s physically impossible for us to hear what’s going on inside the box. as others have already explained, audio waveforms are recreated perfectly up to half the Nyquist frequency. the only audible differences are introduced by other factors (anything randomly introduced each time a file is bounced, aliasing, and the like).

as for why people shouldn’t just take others at their word, that seems a silly argument, when there ways available to objectively prove one way or another whether there’s a difference. that’s to ask “why prove it when you can take it on faith?”

[visheshl]

But I'm gravitating towards 48khz, I'm wondering if 48k does indeed make aliasing more subtle, anything above 48khz is simply making the computer do extra work for nothing. That's just my preference though

[orthodox]

I'm wondering if 48k does indeed make aliasing more subtle

It does indeed, for square or saw harmonics - by 1.5 dB

Actually, 48k is twice more CPU-efficient than 44.1k in case oversampling and anti-aliasing filtering is used by some device. 44.1k requires a steeper brick-wall filter near the Nyquist frequency and so it consumes more CPU cycles.

[visheshl]

I'm wondering if 48k does indeed make aliasing more subtle

It does indeed, for square or saw harmonics - by 1.5 dB

Actually, 48k is twice more CPU-efficient than 44.1k in case oversampling and anti-aliasing filtering is used by some device. 44.1k requires a steeper brick-wall filter near the Nyquist frequency and so it consumes more CPU cycles.

Nice, that's good info. Thanks

[jam-s]

may i say something?

higher hertz and bit depth means moer smooths signal when converting from digital to analog -> to speakers.

if i listen to a decent amount of time let's say all the time at 44,1/16.... let's say for a period of ... i don't know... 5 years?... then when i switched to true 192/24 when i bought external usb sound card... WOOOOOOOOOOW it sounds so different. more rich sound, more smooth sound, more detailed sound.

Well, you spent money on it so it has to sound better, right? Funny thing how our brains twist our perception, right? I suppose it's placebo effect in combination with confirmation bias that you're experiencing with the higher sample rates. of course doing a proper double blind test on this is not that easy, but there are some tools that can help with this.
Be warned thou: There are some really expensive Hi-Fidelity pro-audio outlets (or rather scammers) out there looking for your money and trying to sell you the craziest kind of cables, amps and speakers for ridiculous amounts of cash using some fake science mumbo-jumbo.

[SoundObjects]

Well, you spent money on it so it has to sound better, right? Funny thing how our brains twist our perception, right? I suppose it's placebo effect in combination with confirmation bias that you're experiencing with the higher sample rates. of course doing a proper double blind test on this is not that easy, but there are some tools that can help with this.
Be warned thou: There are some really expensive Hi-Fidelity pro-audio outlets (or rather scammers) out there looking for your money and trying to sell you the craziest kind of cables, amps and speakers for ridiculous amounts of cash using some fake science mumbo-jumbo.

So right - same with speakers.
Many blind test have been made and the result is total different from when the listener know what brand is playing.

[Yonatan]

My ears have suffered too much to even bother any higher than 44kHz. But I do try to go to 48kHz now and then just by mere superstition. I believe that what I cannot hear, someone else might hear.

[avasopht]

It might be better to test the following:
1. Rendering to 44.1khz vs rendering to 48khz and then resampling to 44.1khz
2. Rendering to 48khz vs rendering to 44.1khz and then resampling to 48khz

You can take it a step further by seeing if you get a different sound after running each of those four options through a number of effects (maybe there are subtle artefacts that affect certain types of effects).

If resampling between sample rates introduces notable artefacts (but doesn't the other way around), that would indicate that one sample rate might be better to use given it will likely have to be resampled for each medium (unless you simply export at those sample rates).

[Carpainter]

My ears have suffered too much to even bother any higher than 44kHz. But I do try to go to 48kHz now and then just by mere superstition. I believe that what I cannot hear, someone else might hear.

I do the opposite. Whatever I can't hear very well gets rolled off, which makes even 44kHz overkill for me. I run Reason at a higher sample rate only because the older devices, which I can't seem to stop using, benefit from it.

[enossified]

I'm still at 44.1/24, no reason to change.

Spotify promised lossless (not even hi-res, just not mp3 compression) years ago and it never happened. Where will they get all the content, are they going to upgrade everything currently available or will it only be new pop releases which I don't even listen to. I'm not holding my breath.

I've had so many issues with Spotify, I'll be amazed if rolling out high-res doesn't break all the devices I use for Spotify today. It stopped working properly on my iPod a couple of years ago and I've had numerous issues with Spotify Connect on my stereo system.

[bitley]

If I could use it at 28 or 32 khz I would. I love a little degradation, it creates those warm frequencies I love.