Does 44.1k matter or should I do 96k?

  • Thread starter Thread starter DAS19
  • Start date Start date
How to make a good recording, in order of importance:
1. Good song with good musicians
2. Good room in which to record
3. Good mic'ing technique
4. Good recording equipment
5. Good effects usage


At least this is how I see it. People in general will immediately recognize a strong song played musically, but they will not be impressed with a crappy song that (sonically) sounds amazing.
 
24/44.1
OScope charts do show the improvement, but all the consumer user decks are mainly 16/44.1 quality levels playback. I can't hear a difference from 16/44.1 actually.

it gets to be like these HDTV's...yeah 1080P then 1080i this week...
last week it was 720i, then 720p or vice versus.

most people can't tell with their eyes, unless they read the spec.
the room, the setup of the equipment (very subjective) all plays a such a huge/intregal part.

HDTV is great, but can most tell between 480,720,1080? The OScope can, no doubt there. but 98% of the playback media isn't 1080...and most houses aren't a $50,000 Home Theater room..........and setting the Tones, Brightness etc...is extremely subjective to what the personal user likes.

so we're pretty much screwed as a species is all I'm saying. :p

chase the tech rabbit.....
 
Conventional wisdom is bit rate (16 vs. 24) is more important than sample rate.

Go to 24/48 for the biggest bang for the buck.
 
Another thing not always understood is the subjective loudness of 20khz or even 15khz depending on the individual is way lower than say 1khz, if we can hear 20khz at all.
That's just the nature of human hearing. It's far from flat.

Try it yourself, sending test tones through your monitors with 1khz as your reference. Dont cheat by turning the 20khz up! No there's probably nothing wrong with your hearing. It's normal.

So when people say they can hear the harmonics of 20khz you have to wonder.

Tim
 
The best reason to go with 96 or 192KHz is to line the pockets of manufacturers. I am sure that they would appreciate it!
 
I doubt we can hear beyond 20KHz, but I think it's possible that the frequencies beyond 20KHz can influence the frequencies we CAN hear in subtle ways (sympathetic vibration, beat frequencies, whatever).

But I could see how converting files from one sample rate to another could introduce some unwanted 'stuff'. If you're increasing the sample rate, the computer basically has to invent some data points. Depending on how well the conversion algorithm does that you would get different results. And when decreasing the sample rate...which data points do you get rid of? Different schemes might give different results.

But I've never bothered with anything higher than 24/48. I probably can't hear the difference.


CP
 
Mindset - I think you have a really good point about picture resolution - you can get a really good digital picture at only 3.2 mega pixels if the camera has excellent color representation. The same thing can be said about sample rate ... it's not the rate that matters (when you get to the difference between 44.1 and 48 or 88) but rather what the device is able to capture, or in this case, convert.

If the device is noted for it's ability to capture and convert really well, then it's obviously going to be beneficial to make the jump if everything else warrants it.

But if you have bad/low end pres and/or AD converters, recording at whatever khz is still only going to be as good as the hardware that does the converting and/or capturing. in the case of sample rate, the worst piece(s) of hardware you have for capturing/converting will be what makes the difference between "sounds naturally awesome" and "yuck" if *everything else* is on the money.

I've heard a guitar captured to 16-bit 44khz dat that sounds better than anything I can record at 88.2khz or 192khz ... so it doesn't make sense for me to record at high sample rates since I can get basically the same sound from 44.1, and my hardware seems to work, and sound it's best at it's optimal performance at that sample rate. not to mention that my computer is far more efficient at 44.1. I'll never compromise to 16-bit though ...
 
I attached a simple picture of what I just drew to show some of you an easier representation. There's a reason why Nyquist did what he did. Even at 96khz, that doesn't exactly mean that it's capturing sound from 20khz to 96khz, it just means it's capturing more pictures between whatever sound it could be, maybe between 50hz & 80khz, or 200hz to 1k. They added the extra 2-3 for alias problems that arised, but yeah, the better a/d converters, the better it is, but everyone knows that.
 

Attachments

  • untitled.webp
    untitled.webp
    12.3 KB · Views: 159
crankypants said:
I doubt we can hear beyond 20KHz, but I think it's possible that the frequencies beyond 20KHz can influence the frequencies we CAN hear in subtle ways (sympathetic vibration, beat frequencies, whatever).

But I could see how converting files from one sample rate to another could introduce some unwanted 'stuff'. If you're increasing the sample rate, the computer basically has to invent some data points. Depending on how well the conversion algorithm does that you would get different results. And when decreasing the sample rate...which data points do you get rid of? Different schemes might give different results.

But I've never bothered with anything higher than 24/48. I probably can't hear the difference.


CP
Click to expand...

I can't hear the 17khz ringtone, my kids can. Its the one the teachers can't hear..

try it for fun.
http://www.jetcityorange.com/MosquitoRingtone.html

I even checked it thru my EMU 24/192 setup and the mixer was showing it, but I wasn't hearing it.
 
Mindset said:
I attached a simple picture of what I just drew to show some of you an easier representation. There's a reason why Nyquist did what he did. Even at 96khz, that doesn't exactly mean that it's capturing sound from 20khz to 96khz, it just means it's capturing more pictures between whatever sound it could be, maybe between 50hz & 80khz, or 200hz to 1k. They added the extra 2-3 for alias problems that arised, but yeah, the better a/d converters, the better it is, but everyone knows that.
Click to expand...


Haha, that is really misleading. Generate a 20kHz sine wave at 44.1/16, play it out of your soundcard's DA, and into an O-scope. I can guarantee you that you will see a very normal sine wave (perhaps a slight attenuation from the original amplitude due to the filters being so near with 44.1k), and not the jaggedy piece of crap that you show.
 
Reggie said:
Haha, that is really misleading. Generate a 20kHz sine wave at 44.1/16, play it out of your soundcard's DA, and into an O-scope. I can guarantee you that you will see a very normal sine wave (perhaps a slight attenuation from the original amplitude due to the filters being so near with 44.1k), and not the jaggedy piece of crap that you show.
Click to expand...


It just illustrates that a lower sample rate means less samples are captured per second to the hard drive. But you are right, this question of sample rate can't be answered by theory alone ... it has to be answered practically ... so get out the system, and record some stuff at 44.1khz, 48khz, 88.2khz and 192khz ... see which sounds the best and by how much does it sound the best. If you end up being faced with "higher sample rate "X" sounds better", then ask yourself if it's worth moving up and consider any baggage the move will bring (higher CPU, higher HD space being eaten) ... it might very well be worth it for some.
 
Quote from The Flaming Lips about sample rates:

Mixing and mastering the album brought Michael a useful and unexpected piece of wisdom. “We do all our recordings in high resolution, 24 bits, 88.2kHz,” he says. “When we export tracks, it seems like it’s easier on the computer to do the half calculation from 88.2kHz to 44.1kHz, as opposed to 96 down to 44 or whatever. So we thought, ‘We’re knocking down the bit depth and we’ve heard about this thing called dithering. Why don’t we do an experiment?’ For some reason, [every type of dithering we tried] affected the music and made it sound different. It was as though we put a BBE Aural Exciter on it. So when we export, we just do a simple rate conversion instead of doing the dither thing.”

Quotes from the Crosstudio about sample rates:

1)
The fact that The Flaming Lips believe that they are burning CD's at 24/44.1 [reducing sample rate, but not dithering] almost made me not quote their dumb asses, but it just goes to show that people can be half right and really f*$@ you up.

2)
Mixing individually recorded 44.1khz tracks to a 192khz stereo mix is like having a wallet size photo and taking a picture of it with your cool new 10 mega pixel camera and then being pissed that the picture doesn't look clearer than the original.

3)
If you want a real test.

Record all of your tracks at 24/88.2, do all of your processing in 24/88.2, then burn your CD at 16/44.1.

Now do it all again in 24/44.1 using the same equipment.
 
COOLCAT said:
chase the tech rabbit.....
Click to expand...


Been doing that for years... Especially when it comes to computers. Upgrade even though the version of whatever software it is works fine. If everyone got smart and stopped falling for the hype the world would be a better place.
 
Mindset said:
I attached a simple picture of what I just drew to show some of you an easier representation. There's a reason why Nyquist did what he did. Even at 96khz, that doesn't exactly mean that it's capturing sound from 20khz to 96khz, it just means it's capturing more pictures between whatever sound it could be, maybe between 50hz & 80khz, or 200hz to 1k. They added the extra 2-3 for alias problems that arised, but yeah, the better a/d converters, the better it is, but everyone knows that.
Click to expand...


The whole reason behind higher sample rates is trying to do something digtial was never intended to to do which is be linear like tape. Digital no matter how high the word length is choppy (squared may be a better word to use). Which is why they invented dithering to blur the image. The reason Nyqusit doubled the frequency at which a human could hear was to accurately capture a sample.
 
I think that if recording equipment had not progressed much past what was available in 1959, really good artists would have records 'sounding' about as they do today........ or maybe better.
 
crosstudio said:
Quote from The Flaming Lips about sample rates:

...
3)
If you want a real test.

Record all of your tracks at 24/88.2, do all of your processing in 24/88.2, then burn your CD at 16/44.1.

Now do it all again in 24/44.1 using the same equipment.
Click to expand...


I agree with you on that one crosstudio. :D.

If you're gonna work with higher frequencies you might as well start with in the beginning and work it to the very end. I think that is truly the only way to benefit from higher sampling rates otherwise there really no point in trying to do it at the very end of the signal chain.

Although everything still comes down to cost to do everything well. Perhaps there would be no need for mastering engineers if we had all the best equipment. Unfortunely this is not so either since an experience operator is better than one without any. So better is better!! No surprises there.

But I still think 24 bit / 44.1 khz adequate for most types of music today. I think the most important part of the equation now and always is to make sure you haven't overlook any obvious weak point in the recording chain as southside Glenn pointed out earlier in the thread.

Start your way from the beginning and work it to the end. I know that a seems like pretty obvious statement, but I think people tend to forgot that after a while.


The more you know the better it's gonna get!! That's my philosophy and I still have a lot to learn.
 
bigwillz24 said:
The whole reason behind higher sample rates is trying to do something digtial was never intended to to do which is be linear like tape. Digital no matter how high the word length is choppy (squared may be a better word to use). Which is why they invented dithering to blur the image. The reason Nyqusit doubled the frequency at which a human could hear was to accurately capture a sample.
Click to expand...

Um...Sample rate has nothing to do with making the wordlength more linear. It has to do with bandwidth. People THINK that it will make it more like tape. Although you are right that dither blurs the sound of the least significant bit turning on and off to digital black. Higher bit-depths just push this least significant (quantization bit) bit down further under the noisefloor.
I would also argue that it is the non-linearities of tape that digital encoding was not intended for.

I am curious, however, at what bit depth digital would be equal in "choppiness" to the iron oxide molecules aligned on tape...hmm...
 
DAS19 said:
I was wondering if the difference in quality was really a difference and if I should recording in 96khz. Will it bulk my recordings up becuase they sound kinda dry and lofi on 41.1k

Thanks
Dave
Click to expand...

44,48,96 is good if your converters were optimized in the design stage for anyone of these. You will notice a far bigger difference between 16 and 24 bit at any of these sampling rates. I have heard all of the above sound better or worse. I run 48K 24 bit at tracking all the time because my HD recorder sounds best there. I tried 96K and did not hear a difference. I like my HD recorder to spend less time crunching useless data. If it sounded better at 96K, I would use it. AT mixdown. I go through a big analog mixer straight to a wavefile. Best way in my opinion as you avoid all of the downsampling problems.
 
You must log in or register to reply here.
Back
Top