Quantcast
What sample rate do you record at? - Page 6 - Gearspace.com
The No.1 Website for Pro Audio
What sample rate do you record at?
View Poll Results: What sample rate do you record at?
Record at 44
542 Votes - 43.15%
Record at 48
344 Votes - 27.39%
Record at 88
141 Votes - 11.23%
Record at 96
191 Votes - 15.21%
Depending on the project, I would record at 44 or 96.
108 Votes - 8.60%
Multiple Choice Poll. Voters: 1256. You may not vote on this poll

Old 16th August 2010 | Show parent
  #151
Lives for gear
 
phas3d's Avatar
 
🎧 10 years
Quote:
Originally Posted by Bob Olhsson ➑️
The biggest improvement for me has always been between 44.1 and 48. I actually prefer 96k and even 48k to 88.2 because most of the digital signal processing I have sounds cleaner provided the audio was recorded at those sample rates in the first place. My results from up-sampling have been pretty mixed.
You have no advantage whatsoever in upsampling. If you want to work at higher sample rates you need to start the project at that setting and record and go through everything at the value you've chosen. It's like having an mp3 file and converting it to WAV. It will still have the quality of mp3. Of course you can eq and compress it, but the source is an mp3 file.
Old 16th August 2010 | Show parent
  #152
Motown legend
 
Bob Olhsson's Avatar
 
🎧 15 years
Quote:
Originally Posted by phas3d ➑️
You have no advantage whatsoever in upsampling...
This isn't true in the case of non-linear processes such as limiting. It becomes a question of the trade-off between the distortion created by the up-sampling vs. the added distortion in limiting caused by the low sample rate. That's a subjective call that I've found somewhat unpredictable. The times up-sampling has worked best for me have been when the original recording was made with the highest grade converters. Lower grade seem to produce a more fragile recording.
Old 18th September 2010 | Show parent
  #153
Gear Guru
 
AllAboutTone's Avatar
 
🎧 15 years
I have used 44 and 48, seems like I prefer 48, hated 88, I would like to try 96 sometime.
Old 18th September 2010 | Show parent
  #154
Lives for gear
 
waxx's Avatar
 
🎧 15 years
if it's an ITB project (wich is mostly the case) i record on 44.1. If i'm going to mix it trough a console, the tracking is done in 96 and the master goes to 44.1.

if i do post production (wich is actually already a long time ago and very seldom) i go for 48

I always tries to stay close to the sample rate of the targeted distribution media because i don't like sample rate conversion. So for music it's 44.1 and postproduction 48. I don't work on tape (i know nothing about tape and the machines involved so i stay out.)
Old 18th September 2010 | Show parent
  #155
Gear Head
 
Wild Sky's Avatar
 
🎧 10 years
Time is of the essence

I'd like to add something to the mix.....

Most people here debate about the frequency content of sampling rates, do we really hear it, etc.

IMO, quality of sound is not so much perceived by frequency content as much as time definition. When you record at higher sampling rates, each timing information is better represented. If I need to chose between let's say 150 samples to capture the instant the stick hits the snare or the pick strikes the string, and 300 samples, I'll choose 300!

I also find that if time definition is better, so is depth of field ( you suddenly hear people moving on chair, mic placement is even more obvious, room sound definition is increased...) and definitely stereo image! Left/right and front/back definition at 96 is way better than at 48... YMMV!

I find that when I record at 96, I can mic a little further from the source to get more vibe and it doesn't sound "smeared".

BTW I record at 96 and mix OTB to tape...
Old 18th September 2010 | Show parent
  #156
Gear Maniac
 
🎧 10 years
Talking

Quote:
Originally Posted by ISedlacek ➑️
96 kHz /16 bit heh
LOL heh

---

Thomas Johansen Record Producer
Old 18th September 2010
  #157
Lives for gear
 
bassjam's Avatar
 
🎧 10 years
What sample rate do you record at?

DXD sounds amazing! 384k. Go see the DAD guys at AES and have a listern to some of there recordings.

In the real world we believe that 96/24 is optimum, Daniel Weiss shares this view. When mixing otb with an analogue board the sound is closer to the sound when tracking.
Old 18th September 2010 | Show parent
  #158
Gear Addict
 
Fabi's Avatar
 
🎧 10 years
I choose between 44.1 or 88.2 depending on the project which is not an option so I chose 44.1 or 96
Old 19th September 2010 | Show parent
  #159
Lives for gear
 
Ken Lewis's Avatar
 
🎧 10 years
I havent read this entire long thread, but i've read a few pages. Has anyone at all mentioned that your DSP processing power is cut in half when you double the sample rate? So my HD Accel 5 system just became an Accel 2and a half at 96K. Any session with a significant track count is going to make me feel like my hands are cuffed behind my back at 96K always having to keep an eye on how much processing power i have left, and should i use THIS plugin which is already a DSP hog, or should i use THAT plugin which sounds almost as good but wont tax my system so much.

When i mix i like to listen to music, not worry about whether my DSP is going to run out before i'm done mixing and i'll have to start making compromises. 44.1 to me is WAY less of a potential audio compromise than the above scenario and running out of DSP power late in a mix, or worse, in the middle of a mix.
Old 19th September 2010
  #160
Gear Addict
 
ajv205's Avatar
 
🎧 10 years
I keep it at 24/48 for a few reasons.
- Pro Tools hardware sounds better at 48 than 88.2 to me.
- audio for picture deliverables are still required at 16/48.
- the small improvement in high end does not justify losing very important track count and much larger session file sizes.
- like Ken said, its all about the mix, and higher res recordings do not translate to improved mixes if you still do the same things.

As long as converter technology keeps improving I will stay at 24/48 for quite some time.
Old 19th September 2010 | Show parent
  #161
Gear Addict
 
Johnny Paez's Avatar
 
🎧 10 years
88.2 works for me
thumbsup
Old 19th September 2010 | Show parent
  #162
Lives for gear
 
The Beatsmith's Avatar
 
🎧 15 years
i would be at 88.2 if my plugins (especially arturia minimoog V and vember surge) weren't so buggy at that sample rate. some of my projects are at 88.2, but i've dropped back down to 44.1 just because of all the headaches.
Old 19th September 2010 | Show parent
  #163
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by Wild Sky ➑️
I'd like to add something to the mix.....

Most people here debate about the frequency content of sampling rates, do we really hear it, etc.

IMO, quality of sound is not so much perceived by frequency content as much as time definition. When you record at higher sampling rates, each timing information is better represented. If I need to chose between let's say 150 samples to capture the instant the stick hits the snare or the pick strikes the string, and 300 samples, I'll choose 300!

I also find that if time definition is better, so is depth of field ( you suddenly hear people moving on chair, mic placement is even more obvious, room sound definition is increased...) and definitely stereo image! Left/right and front/back definition at 96 is way better than at 48... YMMV!

I find that when I record at 96, I can mic a little further from the source to get more vibe and it doesn't sound "smeared".

BTW I record at 96 and mix OTB to tape...
You're confused about digital sampling theory.
Old 19th September 2010 | Show parent
  #164
Gear Head
 
Wild Sky's Avatar
 
🎧 10 years
Quote:
Originally Posted by timlloyd ➑️
You're confused about digital sampling theory.

Please enlighten me to what finer points I have missed...

Even the term Khz is a time based concept. X amount of samples/second. If I choose a visual analogy, a 24 frame/second camera has less definition than a 60 Fps one... If you compare 1 frame of each camera, they might be very similar in light level , contrast, colors (frequency content) but when you see 60 of them in 1 second compared to 24, the resolution in time is increased. Movements are more fluid. If a fly passes in front of the lens at 24 fps, you might get 2 frames with the fly on it. At 60 fps, you'll get 5 frames with the fly. At 24 fps, you have 2 points of reference. Fly enters right (frame 1) and exits left (frame 2). At 60 fps you can now see that its path was elliptical, or zigzag. Better time resolution.

96,000 little snapshots of a sound story through time (1 second) is more precise than 48,000.

Heck, sound travels at around 343 meters/second. So in effect, 1 sample at 48 has a resolution of about 7 millimeters to capture the sound's journey through space. What happens between those 7 mm is interpolated by the computer. If the computer has to interpolate between every 3.5 mm, the resulting image is closer to the truth.

YMMV
Old 19th September 2010 | Show parent
  #165
Lives for gear
 
🎧 10 years
Digital photography can't be used as a direct analogy for digital PCM sampling, because the data resolution below the Nyquist frequency does not increase as the Nyquist frequency increases. It is the frequency range that increases, not the resolution of those frequencies. Your analogy is flawed semantically, because you're implying different definitions of the word "resolution" for each scenario.

If we disregard physical circuit limitations/design for a moment, recording a 1kHz sine wave at 96kHz does not capture it more accurately than recording it at 44.1kHz. This is mathematically true, according to the theories of Nyquist and Shannon that underpin the very existence of digital audio.

The thing is, frequency content and time definition are inherently linked. An audio transient is a burst of high frequencies relative to the preceding and following frequencies. The stick attack of a snare drum sounds sharp and fast because the attack portion of the sound envelope contains high frequencies, so if frequencies above 20kHz are not directly audible, then in order to completely capture everything of that sound that *you are hearing*, you only need to sample at 44.1kHz.

That's in a perfect theoretical situation though, and in reality there are/can be aspects of circuit design that can/do cause an ADC to sound a lot different at 44.1kHz vs higher sampling rates, but that doesn't change the fact that for digital PCM audio, you don't actually *need* any higher than 44.1kHz at the recording stage. In fact, if you think that Dan Lavry knows what he's talking about , then you would understand that for some ADC circuit designs, high sample rates can actually induce greater error and therefore lower sound quality than they produce at 44.1kHz.

Actually, for standard music production aimed at human consumption, the reason to use a high sample rate vs 44.1kHz is so that the pass-band ripple of the anti-aliasing filter section of the circuit doesn't encroach into the audible frequency range. That's why the theoretically optimum sample rate (according to Lavry) is around 60kHz, because it is sufficiently high to eliminate filter ripple in the audible range, but low enough to minimise error due to the physical limitations of the ADC circuit. Unfortunately though, the market is populated by companies who want to propel the "faster is better" myth, and sell 192kHz-capable gear to people who have been purposefully misinformed about what they're using. Not very many DAWs allow you to record at Lavry's optimum sr, and a lot of plug-ins aren't optimised for that sr, and wouldn't work correctly, so it's not used very often.

Digital processing is a different situation, for which it often does produce much better results at higher sample rates, but that's because of potential aliasing and other Nyquist-effects; *not* because the audio has been captured at greater "resolution", because it hasn't. And all of that depends on the savvy of the software developers.........aliasing and EQ curve distortion can be avoided at lower sample rates.

The analogy of a fly to a sound wave isn't compatible. If a fly could only move bi-directionally in a sinusoidal curve, then you would only need 2 sample points per wavelength to be able to 100% reproduce its flight path at any point in time.........so you would need to decide what the highest wavelength was that you wanted to capture, and then sample at double that. But they don't move in the same way as sound waves.................

So, I'm not saying that recording at higher that 44.1kHz is pointless, I'm just saying that the common justifications for doing so are mis-informed, and just using the highest possible sr that you can isn't wise, or necessarily going to produce the best results. Pulse-density modulation (DSD for SACD production) doesn't work in the same way as PCM, so the justifications for using MHz sample rates are different, as the digital data is stored and reproduced in a different way. Both systems have upsides and downsides.

MMDV
Old 19th September 2010 | Show parent
  #166
Motown legend
 
Bob Olhsson's Avatar
 
🎧 15 years
Quote:
Originally Posted by timlloyd ➑️
You're confused about digital sampling theory.
What does digital sampling theory have to do with digital sampling practice?

Sampling theory is a discussion of potential that totally ignores real-world implementation which is what real world audio practitioners are forced to work with every day. There's more than enough bad math, software bugs and incompetently designed conversion hardware around to make people's subjective observations absolutely valid.
Old 19th September 2010 | Show parent
  #167
Gear Head
 
Wild Sky's Avatar
 
🎧 10 years
Quote:
Originally Posted by timlloyd ➑️
Digital photography can't be used as a direct analogy for digital PCM sampling, because the data resolution below the Nyquist frequency does not increase as the Nyquist frequency increases. It is the frequency range that increases, not the resolution of those frequencies. Your analogy is flawed semantically, because you're implying different definitions of the word "resolution" for each scenario.

If we disregard physical circuit limitations/design for a moment, recording a 1kHz sine wave at 96kHz does not capture it more accurately than recording it at 44.1kHz. This is mathematically true, according to the theories of Nyquist and Shannon that underpin the very existence of digital audio.

The thing is, frequency content and time definition are inherently linked. An audio transient is a burst of high frequencies relative to the preceding and following frequencies. The stick attack of a snare drum sounds sharp and fast because the attack portion of the sound envelope contains high frequencies, so if frequencies above 20kHz are not directly audible, then in order to completely capture everything of that sound that *you are hearing*, you only need to sample at 44.1kHz.

That's in a perfect theoretical situation though, and in reality there are/can be aspects of circuit design that can/do cause an ADC to sound a lot different at 44.1kHz vs higher sampling rates, but that doesn't change the fact that for digital PCM audio, you don't actually *need* any higher than 44.1kHz at the recording stage. In fact, if you think that Dan Lavry knows what he's talking about , then you would understand that for some ADC circuit designs, high sample rates can actually induce greater error and therefore lower sound quality than they produce at 44.1kHz.

Actually, for standard music production aimed at human consumption, the reason to use a high sample rate vs 44.1kHz is so that the pass-band ripple of the anti-aliasing filter section of the circuit doesn't encroach into the audible frequency range. That's why the theoretically optimum sample rate (according to Lavry) is around 60kHz, because it is sufficiently high to eliminate filter ripple in the audible range, but low enough to minimise error due to the physical limitations of the ADC circuit. Unfortunately though, the market is populated by companies who want to propel the "faster is better" myth, and sell 192kHz-capable gear to people who have been purposefully misinformed about what they're using. Not very many DAWs allow you to record at Lavry's optimum sr, and a lot of plug-ins aren't optimised for that sr, and wouldn't work correctly, so it's not used very often.

Digital processing is a different situation, for which it often does produce much better results at higher sample rates, but that's because of potential aliasing and other Nyquist-effects; *not* because the audio has been captured at greater "resolution", because it hasn't. And all of that depends on the savvy of the software developers.........aliasing and EQ curve distortion can be avoided at lower sample rates.

The analogy of a fly to a sound wave isn't compatible. If a fly could only move in a sinusoidal curve of acceleration and deceleration, then you would only need 2 sample points per wavelength to be able to 100% reproduce its flight path at any point in time.........so you would need to decide what the highest frequency was that you wanted to capture, and then sample at double that frequency. But they don't move in the same way as sound waves.................

So, I'm not saying that recording at higher that 44.1kHz is pointless, I'm just saying that the common justifications for doing so are mis-informed, and just using the highest possible sr that you can isn't wise, or necessarily going to produce the best results. Pulse-density modulation (DSD) doesn't work in the same way as PCM, so the justifications for using MHz sample rates are different, as the digital data is stored and reproduced in a different way. Both systems have upsides and downsides.

MMDV
I can't argue the maths... But I know a cymbal or a hihat sounds smoother a 96khz than at 48 (using high quality converters in both cases), even if my conscious hearing range is the same in both SR. And that even if I perceive the same freq response for the snare attack, 96 sounds closer to the actual microphone without any AD/DA. I doesn't sound brighter, it sounds more real. I perceive more depth of field. So maybe there's more to it than our conscious freq range...

Peace,
Old 19th September 2010 | Show parent
  #168
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by Bob Olhsson ➑️
What does digital sampling theory have to do with digital sampling practice?

Sampling theory is a discussion of potential that totally ignores real-world implementation which is what real world audio practitioners are forced to work with every day. There's more than enough bad math, software bugs and incompetently designed conversion hardware around to make people's subjective observations absolutely valid.


I think that depends on the situation; the practice wouldn't even exist if not for the theory, but I know that's not what you're getting at.

Understanding the theory behind processes helps me personally, maybe it's not the same for others. I can't comment on the practices of others here, I can only comment on what I read and infer.........and in this case I saw "mis-information."

Maybe I should go back to the "music computers" sub-forum where I belong


Quote:
Originally Posted by Wild Sky ➑️
So maybe there's more to it than our conscious freq range...
Maybe there is, the extent of it has yet to be determined (yes I know about the Oohashi paper). I may be doing some research on this myself over the next year
Old 19th September 2010 | Show parent
  #169
Gear Head
 
Wild Sky's Avatar
 
🎧 10 years
Quote:
Originally Posted by timlloyd ➑️


Maybe there is, the extent of it has yet to be determined (yes I know about the Oohashi paper). I may be doing some research on this myself over the next year.
Keep us posted, that subject is worth it's own thread!
Old 19th September 2010 | Show parent
  #170
Motown legend
 
Bob Olhsson's Avatar
 
🎧 15 years
Quote:
Originally Posted by timlloyd ➑️
...Understanding the theory behind processes helps me personally, maybe it's not the same for others...
It certainly has helped me however I have a problem with the many suggestions (and I don't mean from you) that people must be delusional if they observe something that contradicts theory when there's always a very real possibility that something is broken.
Old 19th September 2010 | Show parent
  #171
Gear Head
 
Wild Sky's Avatar
 
🎧 10 years
Quote:
Originally Posted by Bob Olhsson ➑️
I have a problem with the many suggestions that people must be delusional if they observe something that contradicts theory when there's always a very real possibility that something is broken.

That's worth a sig line!!
Old 19th September 2010 | Show parent
  #172
Lives for gear
 
DarkSky Media's Avatar
 
🎧 10 years
Quote:
Originally Posted by timlloyd ➑️
...An audio transient is a burst of high frequencies relative to the preceding and following frequencies...
An audio transient is a dynamic event (rapid change in amplitude) not a frequency event.

Transients (as in your snare stick-hit example) are often accompanied by high frequency content, so I do understand what you're saying here, but it's not helpful to merge the two concepts. You can't necessarily infer slew rates from published FR specs for example.
Old 19th September 2010 | Show parent
  #173
Lives for gear
 
🎧 10 years
That's a fair point, although I would probably use the word onset rather than transient if there's not relatively significant high frequency content.

Quote:
Originally Posted by Bob Olhsson ➑️
It certainly has helped me however I have a problem with the many suggestions (and I don't mean from you) that people must be delusional if they observe something that contradicts theory when there's always a very real possibility that something is broken.
I agree, if someone is hearing something, then there is a reason. However, that reason might be a placebo, and I'm always interested in finding out.
Old 19th September 2010 | Show parent
  #174
Motown legend
 
Bob Olhsson's Avatar
 
🎧 15 years
Quote:
Originally Posted by timlloyd ➑️
I agree, if someone is hearing something, then there is a reason. However, that reason might be a placebo, and I'm always interested in finding out.
Most people who have any experience in professional audio are painfully aware of that possibility.

Finding out is interesting however it's not nearly as simple as some would have people believe due to the variations in hearing among individuals and the way our brain filters and steers our attention to what we hear.

My experience has been that where there is smoke, there generally turns out to be fire however it can take a long time to determine exactly what the mechanism involved is.
Old 19th September 2010 | Show parent
  #175
Lives for gear
 
jchadstopherhuez's Avatar
 
🎧 15 years
tried em all.

happy tracking/overdubbing everything at 48k/24 bit on RADAR Classic converters.

mixing often in logic/cubase/pt 8.0 with UAD cards...again...at 48k. print final mixes to 1/2" ATR or 1/4" OTARI analog...then back to a second DAW at 88/24 bit via HEDD or Apogee AD8000 @ 44/24 bit.

works for me.

best,

jchristopherhughes
Old 19th September 2010 | Show parent
  #176
Gear Head
 
🎧 10 years
24bit 88.2k

There really isn't much of a difference between 88 and 96. At least not from my experience. So I record at 88.2 because the coversion back to 44.1 is smoother.
Old 19th September 2010 | Show parent
  #177
Lives for gear
 
laser's Avatar
 
🎧 15 years
Quote:
Originally Posted by Wild Sky ➑️
I can't argue the maths... But I know a cymbal or a hihat sounds smoother a 96khz than at 48 (using high quality converters in both cases), even if my conscious hearing range is the same in both SR. And that even if I perceive the same freq response for the snare attack, 96 sounds closer to the actual microphone without any AD/DA. I doesn't sound brighter, it sounds more real. I perceive more depth of field. So maybe there's more to it than our conscious freq range...

Peace,

I've done countless AB tests on cymbals and other instruments. On every test, cymbals sound better at 96kHz with my converters. I also found that electric guitars with distortion sound slightly better at 48kHz.

So, with my converters, it would depend on if you value the pristine sound of cymbals or the slightly better sound of electric guitars with distortion. I'm sure many people on Gearslutz people would choose the latter.

I split the difference and went 88.2 KHz. heh

Laser
Old 19th September 2010 | Show parent
  #178
Lives for gear
 
🎧 10 years
I don't think that non-integer ratio SRC is any different in quality to integer ratio SRC. I'm going to refrain from writing another essay though thumbsup
Old 20th September 2010 | Show parent
  #179
Gear Maniac
 
El Nino's Avatar
 
🎧 10 years
Quote:
Originally Posted by zoom ➑️
e.g. bats, dog whistles or remote controls of garage doors.
LMFAO!

"The mix sounds great but, can you please bring up the dog whistle 1Db"

In a perfect world the dual rig would be wonderful. 96K in with great A/D - 96K out with great D/A - into an SSL for mixing - and back into another rig with an even better A/D - and off to Sterling. I've done many a record in my 10 years in. All but 2 were done @ 96K. The only cat I've done 44.1K with is Ron St. Germain. That was in 2000 and he is a die hard old school cat. Most of the projects I've done have gone to Sterling NYC for mastering. IMO, therein lies the question: who is mixing and mastering your material & how are they converting back to 44.1? Ultimately, the songs are what will make or break the record. The rest may very well be personal preference. May very well be not.

As a producer/engineer, I feel very comfortable using 96K. IMHO, broader bandwidths could possibly open up certain sounds. Perhaps not everything BUT, cymbals, acoustic guitars, piano, and vocals may very well benefit from higher sample rates. If you are mixing & mastering yourself, well, just use what works best for your work flow. If you are handing it in to be mixed & mastered by high budget cats, you may want to keep the sample rate broad. Not because it will make or break the record but, because your cymbals MAY sound a tiny bit airier at the end of the day. If you're not worried about that then, just do what feels good to you. BTW, this topic is almost like talking about UFO's. heh

The record I'm currently wrapping up was all done @ 96K by special request of Jay Baumgardner. Go figure. Have fun battling it out guys!
Old 20th September 2010 | Show parent
  #180
Lives for gear
 
🎧 15 years
I dont believe, based on my own experience on way too many occasions, that it's a 'placebo effect' that draws me to 96k recording

it just sounds BETTER to me, and this is the only reason to make that decision

but having said that, it's worth pointing out that current thinking in medicine is that the placebo effect is IMPORTANT.
it's an established part of healing.

not only are patients who believe they will get better more likely to get better, but patients whose DOCTORS believe they are likely to get better also are more likely to get better.
Believing in a treatment's efficacy is the first step toward its working.

It's not a substitute for demonstrable, verifiable, proof... but it's an important adjunct.


telling yourself "ah it's all crap. I read it on the internets that sampling rate doesn't matter" is setting yourself up for a certain listening bias every bt as much as "using the highest sample rate must be best" is.
BOTH have potential influence on your actual experience.


If A consistently sounds better to you than B, go with A.
never mind the internets.
that's allegedly why YOU are driving the car.
πŸ“ Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 4616 views: 615182
Avatar for smoke
smoke 7th May 2021
replies: 55 views: 32243
Avatar for IM WHO YOU THINK
IM WHO YOU THINK 13th October 2020
replies: 98 views: 39100
Avatar for dfghdhr
dfghdhr 5th June 2021
replies: 1296 views: 181653
Avatar for heraldo_jones
heraldo_jones 1st February 2016
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump