Bit Depth / Sample Rate

Messages
1,308
So I have been recording at 24/48 because supposedly that is the "studio standard", but if my target finished product is CD quality (or worse yet streaming), is there any advantage to going beyond 16/44.1? My interface is capable of way better at the expense of a little latency, but if I am mostly recording live instruments with decent-but-average mics in a treated-but-not-perfect room, do I stand to gain anything? Thanks in advance.
 

rumbletone

Silver Supporting Member
Messages
8,266
In the late 80s when everything was sold on cassette tapes, was there any reason to track and mix on a higher quality standard? Absolutely!

Others here are more qualified to talk about floating bit processing and dithering etc etc but the bottom line is that yes there are several reasons to track and mix at at least high bit rate, if not also higher sample rate, then bring down to 16/44.1k or mp3 or other formats only at the mastering stage.
 

stevel

Member
Messages
15,614
So I have been recording at 24/48 because supposedly that is the "studio standard", but if my target finished product is CD quality (or worse yet streaming), is there any advantage to going beyond 16/44.1? My interface is capable of way better at the expense of a little latency, but if I am mostly recording live instruments with decent-but-average mics in a treated-but-not-perfect room, do I stand to gain anything? Thanks in advance.

16/44.1 is all you really need. But, recording at 24/48 is becoming standard as "future proofing".

You're going to have to dither if you're mixing down to 16/44.1 for CD.

But one of our publishers actually asked for the 48k versions for their streaming, and luckily I had recorded them there.

Any higher than that and you're really just wasting disk space (but hey, it's cheap enough now, right?).

And anyway, someone's going to listen to an mp3 version of it, on some bluetooth headphones...

Mix for that :)

Honestly, it's really about capturing a great performance of a great song. It all starts there. If the quality is missing a few bits and it's an awesome tune, played well, and recorded in such a way as to not distract from the performance (aside from "made in the DAW kinds of recordings), then it's not a concern IMHO.

I'm going to record a Costco Acoustic with an SM58 but do it at 96k...

I'd say the eggs are in the wrong basket then :)

I think based on what you say, if the final product is a CD, 16/44.1 is fine to avoid worrying about dithering down. However, if you feel it's going to need to really be future proofed, or that you've got tons of disk space and processing power, there's no harm in working in 24/48 and then dithering down for the CD, keeping the master at the original rate.

After all, in the old days, the master tapes were as good as we could get then, but we listened to LPs with scratches, LPs on the radio with static, and cassette tapes with hiss. Everything we heard was already 2nd generation - and there are people out there screaming that that "sounds better"...

So just because we can increase the depth/rate, doesn't mean we should, or doesn't mean that doing so is going to "improve" the final product in any way (Nyquist already proved all this, and recent polls have show young people actually prefer the sound of songs on MP3s, because that's what they first hear everything on).

I'd pick one and go with it, and focus on the playing, the micing, the mixing, and that stuff.
 

KHAN

Member
Messages
4,634
This is just my opinion (based on reading tons and tons of stuff at Gearslutz and other engineering/mixing forums).

On modern digital gear:

Always record at 24 bits because you can set recording levels around -10. This gives you lots of headroom without degrading sound.

16/44.1k is fine as a delivery medium.

Higher sample rates are fine, but not necessary to make good sounding recordings.

96k will use more than twice the processing power when mixing. So keep that in mind if plugin count is an issue.

I suggest doing some experiments with various sample rates (I've never heard anyone disagree with recording at 24 bits). Do some double blind tests with your files and see if you hear the difference. There is a lot of conformation bias out there.
 

sws1

Member
Messages
12,897
96k will use more than twice the processing power when mixing. So keep that in mind if plugin count is an issue.

I think this may NOT be the case if using UAD plugins as their platforms are engineered around 96k. In fact, I think they say there is more latency with their devices when using 44k.
I could be wrong. Maybe my statement is only applicable when recording.

EDIT: Correcting myself. Latency with UAD is better at 96, but it does use alot more of their DSP. Less plugins can be loaded.
 
Last edited:
Messages
2,325
If you are going to distribute at 16/44.1, track and master at 24/88.2. Using a sampling rate that's not an integral multiple (i.e., 48 or 96 kHz) means that there will be a lossy conversion somewhere in the process between mastering and distribution.

That said, generally speaking there is no good reason not to use a higher bit depth or sampling rate.
 

KHAN

Member
Messages
4,634
I think this may NOT be the case if using UAD plugins as their platforms are engineered around 96k. In fact, I think they say there is more latency with their devices when using 44k.
I could be wrong. Maybe my statement is only applicable when recording.
Could be. OP didn't specify specific gear. My suggestions are pretty much generic based on what I could glean from some pretty heated discussions.

Finding a method of working that you're comfortable with is probably more important than the specific details these days. I settled on 24/44.1k and am good with that for my band demos.
 

SolidGuitar

Member
Messages
607
And don't forget that many plugins internally upsample. So, you may record at 24/44.1k and get _some_ of the benefits of a higher sample rate.

(But please correct me if I'm wrong about this!)
 
Messages
1,936
right ^ above - I did a bunch of research and inquiries about this maybe three or four years ago.
24bit vs 16bit - I can hear a difference. It's subtle but definitely there there
however sample rate : can't hear any diff b/t 44.1, 48 & 96
if I were producing for video, I'd go to 48 but any higher rate number eats-up space on my SSD much faster.
24 / 44.1 sounds pretty good to me and I'm a practical person :cool:

peace
 

jmoose

Member
Messages
5,163
This is just my opinion (based on reading tons and tons of stuff at Gearslutz and other engineering/mixing forums).

Ok, but do you have an opinion based on personal experience?

I've done a lot of blind shootouts. No contest for me to track at 88/96kHz whenever possible. Have printed all my mixes there for at least a decade.

Back around '06 I started working on an album at the Phish barn that became Swampadelica's Sac Passe... Plan was track to 2" and put it right in slow tools from the repro head of the tape machine. Pete Carini and I worked out some routing to allow monitoring in real time and dump tracks in one pass.

Before we got rolling we had to setup the DAW and he was insistent on running at 88.2 while I wanted to stay at 44.1 because, at least in those days drive space was a premium. So I said lets do a single take, then dump it at 44.1 and 88.2 and let us pick. If we all hear a difference and the 88.2 sounds better we'll go with that.

We all heard the difference. No contest. The 88.2kHz sounded way closer to the 2" source. Less "pinched" and more depth.

As for mastering & release, most "real mastering" cats have a chain of analog gear and resample on capture at the end of the chain so it really doesn't matter what sample rate you give them.

Source D/A > gear process > Destination A/D = Final master

Best thing I could advise is to try things yourself. Next time you start a fresh project go for the higher sample rate. See if you don't think it sounds better yourself. If things like EQ and reverbs have more depth and are easier to dial in... less ear fatigue and so on. If you don't find there's a difference over the course of the project go back. No blood no foul. But maybe, just possibly you'll find yourself enjoying the process more at a higher rate and wind up with a better sounding product.

After all that is what the whole recording thing is about...
 

Larry Eh?

Member
Messages
874
Ok, but do you have an opinion based on personal experience?

I've done a lot of blind shootouts. No contest for me to track at 88/96kHz whenever possible. Have printed all my mixes there for at least a decade.

Back around '06 I started working on an album at the Phish barn that became Swampadelica's Sac Passe... Plan was track to 2" and put it right in slow tools from the repro head of the tape machine. Pete Carini and I worked out some routing to allow monitoring in real time and dump tracks in one pass.

Before we got rolling we had to setup the DAW and he was insistent on running at 88.2 while I wanted to stay at 44.1 because, at least in those days drive space was a premium. So I said lets do a single take, then dump it at 44.1 and 88.2 and let us pick. If we all hear a difference and the 88.2 sounds better we'll go with that.

We all heard the difference. No contest. The 88.2kHz sounded way closer to the 2" source. Less "pinched" and more depth.

As for mastering & release, most "real mastering" cats have a chain of analog gear and resample on capture at the end of the chain so it really doesn't matter what sample rate you give them.

Source D/A > gear process > Destination A/D = Final master

Best thing I could advise is to try things yourself. Next time you start a fresh project go for the higher sample rate. See if you don't think it sounds better yourself. If things like EQ and reverbs have more depth and are easier to dial in... less ear fatigue and so on. If you don't find there's a difference over the course of the project go back. No blood no foul. But maybe, just possibly you'll find yourself enjoying the process more at a higher rate and wind up with a better sounding product.

After all that is what the whole recording thing is about...
It would have to be a double blind test, though. Otherwise, it would all mean absolutely nothing.
 

alivegy

Member
Messages
1,176
In my experience the difference is hardware dependent. Bit depth is the amount of information in each sample and basically sets your headroom and noise floor. There isnt much of a benefit in going above 24 bits. Sample rate sets the max frequency that you can capture. You need double the sample rate of a given frequency to accurately capture it. Having a higher sample rate does not give you more detail or information at a given frequency, it doesnt work that way. At 44.1 you have very little wiggle room between the max frequency that you can capture and the audible range. This means that you have to have a sharp shelf to cut off the frequencies above 22 khz. This sharp shelf, especially in cheaper equipment can introduce artifacts, and these artifacts can be exaggerated once you run the results through some plugins. Recording at a higher sample rate allows there to be a gentler slope and that cut off is even farther out of the audible range with less possibility of the noise influencing the audible spectrum.

The best thing that you can do is run a test at each level and do a blind test to see if you can hear the difference. I currently use 24/48. There was a noticeable difference in quality between 44.1 and 48 in the higher frequencies when i started introducing eq and pushing things around.
 

Shiny_Beast

Gold Supporting Member
Messages
11,423
Some plugins might work better at high sample rates, although I bet most of them upsample as required these days.

Call me old fashioned, but 48-44.1 down sampling still makes me nervous.
 
Messages
3,349
I tried to post this last night...but had problems with my mobile browser.

First thing you need to know is that modern sample rate conversion is really quite excellent...it’s not like the early days of digital, where it introduced artifacts for non integer divisions of sample rate (e.g. 48khz to 44.1 as opposed to 88.2 kHz to 44.1 kHz).

Recording at higher bit depth gives you more headroom at the expense of a little bit more storage space requirement (50% more). It’s worth it...just do it!

Recording at higher sample rate reduces the work that anti-alias filters need to do at the expense of significantly more storage (at least 100% more than 44.1 kHz),and more CPU requirements. Try it for yourself to see if it’s the right thing for you. Yes, as mentioned above the latency can be lower (a 32 sample buffer at 96k is half the latency of a 32 sample buffer at 48k)...but you need a great CPU to pull it off. Another place where the higher sample rate can be of benefit is with soft-synths...since not all of them have great anti-alias filtering; and many of them generate high order harmonics from saturation or other wave-shaping processes.
 




Top Bottom