16 bit/44.1 khz versus 24 bit/96 khz.. big difference?

[armanHammer]

The reason is, I'm looking at the Zoom H4 (higher resolution) versus the Foxtex MR8HD (lower resolution, but more tracks, 40gig hard drive, and two more xlr inputs)

so is there a big difference in sound quality?

[cochese]

The big difference between 16 bit and 24 bit is headroom. I'm not familiar with the two units you mention and I never record above 44.1 because I find the the bit depth to make the biggest differnce. With rock and pop music there is not a big dynamic range like you would encounter in a classical recording.

Recording at 24 bit will allow you to record hotter signals and have much less a chance of digital distortion assuming your A/D convertors are good.

[AndrewSimon]

24 bit is a must in my book.

16bit = 65, 536 step resolution
24bit = 16,777,216 step resolution

44.1 KHz is OK
96 KHz might give you a slightly better response in the ultra high
frequencies (guitars don't go there anyway)

I always record in 24bit/44.1KHz

[drfrankencopter]

It doesn't make THAT much difference. I've done the tests, and if you know what you're doing, a 16 bit converter is just fine. Real 24 bit converters are hard to find, most offer about 20 bit performance (which is still pretty damn good), and expensive ones offer 22 bit performance. The rest of the bits are noise...FWIW, 24 bits can resolve the 1/f noise of a 150 ohm resistor when operated at standard audio voltages (just to give an idea of just how much resolution 24 bits is). Good wide format analog decks had S/N around 60-70 dB, which was then improved upon with noise reduction. 16 bit gives you 96 dB resolution...better than any tape deck.

Internal bit depth matters though...calculations should be done in at least 48 bit integer or 32 bit floating point, to give you the headrooms for mixing.

96 Khz....totally not worthwhile IMO. A big waste of disk space and processing power for energy that is out of the audible band.

Cheers

Kris

[MichaelK]

96 Khz....totally not worthwhile IMO. A big waste of disk space and processing power for energy that is out of the audible band.

I agree, unless one has access to very high end D/A and analog mixing, which most home studios don't.

No reason you can't make great demo-quality recordings with 16-bit, maybe even master quality. But there might be a difference in sound quality between the two devices anyway, having nothing to do with bit depth or A/D conversion at all. Unfortunately I'm not familiar with either of them.

[clothwiring]

I always record in 24bit/44.1KHz

Same here.

[countandduke]

I hear a difference with my presonus Firepod running at 24bit and 96khz. I recorded with a Roland vs-1680 for years and that ran at 16 bit. I think the main thing I hear is overall better clarity but someone might explain this as better A/D converters and better mic pres on the Presonus side. Those Firepods are like $400 now!!!! I LOVE mine and will probably pick up another one soon.

Chris

[andybaylor]

24/96 sounds right to my ears. (RME Fireface 800)

I'll never go back.

[LSchefman]

I generally work at whatever bit length and sample rate clients request. In the TV ad world around here, that's usually 24/48.

[Long2Play]

The difference between 24 bit and 16 bit audio is not in how loud the levels can go. Digital Zero is Digital Zero. The difference is in the resolution of the fine harmonic details that exist at the smallest amplitudes. 24 Bit has 256 times more amplitude and dynamic resolution than 16 bits, from a math point of view. 2 to the 16th, vs. 2 to the 24th. Huge and WAY audible difference. The best 16 bit converters have 90-92dB of dynamic range and the best 24 bit converters are around 115dB. Huge difference.

Bit Depth is dynamic resolution and Sample Rate is frequency resolution. Much more is gained via Bit Depth.

[cochese]

I hear a difference with my presonus Firepod running at 24bit and 96khz. I recorded with a Roland vs-1680 for years and that ran at 16 bit. I think the main thing I hear is overall better clarity but someone might explain this as better A/D converters and better mic pres on the Presonus side. Those Firepods are like $400 now!!!! I LOVE mine and will probably pick up another one soon.

ChrisActually I beieve the Roland-VS1680 was 20 bit. The original VS880 was 16 ( I had both ). In the case of the Presonus gear over the Roland many companys use the same convertors and the difference you are hearing has more to do with the cheap micpres that come with these units. The 1680 had very noisy mic and line-in pres. The Presonus gear is much quiter and their preamps are pretty decent considering the price point.

As for the difference between 16/24 bit for a beginner the difference would probably be very hard to hear mainly because the beginner doesn't have all the other pro gear to hear the difference. Expensive monitors, treated room, high end mic pres and compressors etc...

My advice would be to decide how serious you are about recording before taking the plunge on high or low end gear. I tried to help a friend who bought several thousand dollars worth of recording gear get up and running. In short he did not have the patience to take the time to learn how to use all this great gear for what is basically a very part time hobby. He would have been better off with Garage Band.

[Dave Klausner]

Definitely record at 24 bit since it gives you a few advantages. First off is the low level resolution. Even if you are going to make a 16 bit CD at the end, it's better to be doing math with the higher resolution words. Since digital audio is essentially doing math all day, here's an analogy.

Imagine that a 16 bit word is a whole number (integer) and that a 24 bit word has one digit after the decimal point (there's actually a lot more resolution than that, but let's keep it easy). Let's say that all we are doing is adding up 3 numbers (obviously, we are doing more complicated math in digital audio, and a lot more of it!).

1.3 +
1.3 +
1.3
_____
3.9

So in the 24 bit world, we would get an answer of 3.9. When we rounded it off to make our 16 bit CD, we would get an answer of 4. In a 16 bit recording, we would round off each number on the way in, so we would have:

1 +
1 +
1
____
3

So in the 16 bit world, our final answer that we put on the CD is 3. The answer of 4 that we got using 24 bit calculations is much closer to the original 3.9 that we were trying to reproduce. Multiply that by countless thousands of calculations, and it does begin to make a difference.

Headroom is the other big factor. In the analog world, we recorded hot since tape had a certain amount of noise, and to get the best S/N ratio, we hit the tape hard. In digital recording, the rule of thumb is that every 6 dB down is 1 bit less resolution, so in 16 bit recording, we really have to push toward zero to maintain our resolution. If your peaks are at -12 dB, you are making a 14 bit recording, which is quite noticeable.

In a 24 bit recording, you have a ton of extra headroom. You are doing really well if you have 20 bits of real resolution in your system (above the noise floor of your room, mics, preamps, etc.). That means that your peaks can be at -24 dB, and you are still capturing every bit of resolution available to your system. Not only does that mean that you can set levels where you don't have to be afraid of clipping, but you can record your files at lower levels.

The advantage of recording at lower levels means that you have that much more headroom to play with. Some plug ins automatically add gain to the system. Even if they don't, how many times have you added a 6 dB boost to a frequency on an EQ? It's really common to either clip plug ins without knowing, or overload the mix bus by recording hot. That's one of the things that contributes to a "small" sounding digital mix.

On an analog deck, I certainly wouldn't try to record at +18 dB - not only would it distort the tape, but I would have no headroom left for processing or even summing. Well, digital zero is not the same as 0dB in the analog world. In my rig, digital zero is calibrated to -18 dB, so if I recorded at 0 dB (or as close as I could come), I am recording at the equivalent of +18 dB in the analog world.

Regarding sample rates, theoretically, the only thing you are getting with higher sample rates is the ability to reproduce higher frequencies, which we can't hear. In practice, some converters will sound better at higher sample rates, since the slope of the antialiasing filters is more gentle, and it's easier to make a gentle filter stay linear than a steep one. Also, some processes involving distortion will sound better at higher sample rates because of how the math is done. That's not just amp simulators, but also vintage compressor emulations and many other plug ins. Having said that, the differences to my ears are slight, and the bandwidth higher sample rates eat up is not worth the DSP hit to me. I'll make better recordings at 44.1 kHz with all my processing than I will at 96 kHz with less than half the processing.

[Figher53]

If you're making recordings for you and your friends and family (which is what I would assume given the pieces of equipment you're choosing between), I wouldn't worry too much about it. Get the one you're most comfortable with and have some fun.

[s2amps]

I usually run at 24/88.2. I don't use 96 because at 88.2, the processor divides by 2 to get down to 44.1 instead of 2.17687 when mixed down to 16/44.1 for a CD. That floating point math uses a lot of CPU cycles and can introduce artifacts.

[Dave Klausner]

While that seems like a logical way of thinking, in actuality, nobody these days uses that sort of "direct integer" sample rate converter. All modern systems upsample everything to about 2.7 MHz, and use decimation filters to get back down to the desired rate. This is very similar to what oversampling A/D converters do, and the artifacts are similar to what you find from that part of the conversion process, which is to say, they are present but not very noticeable at all. You won't get any more artifacts when converting from 96 kHz to 44.1 kHz than you will going from 88.2.

[KungFuLio]

96k/24bit mix to 1/4" 30 ips when ever i can.

unfortunately, clients budget usually only allow me 44.1/24bit

it's ok but 96/24 bit and great converters are the way to go for me

[atticus]

Keep in mind you will have lower conversion latency at a higher sample rate. It's not just the extended frequency response that makes higher sample rates more desirable.

[drfrankencopter]

Keep in mind you will have lower conversion latency at a higher sample rate. It's not just the extended frequency response that makes higher sample rates more desirable.

True, the converter latency goes down, but overall latency might increase in a DAW system as buffers may need to be increased to handle the extra disk/processor load. Additionally, if latency is to be lower on a high sample rate converter, then the anti-alias filter cannot be much smoother (less slope...a common argument for choosing higher sample rates) than a lower sample rate converter, it's just higher up in the frequency spectrum. It's the digital filters that are responsible for the latency (right?), given today's massively oversampled converters...sample time delay should be small, but a finite # of samples will be required for the filter, and will ultimately drive the latency.

Here's a question for you atticus, do you (or converter manufacturers in general) design with different DSP filters for 16 bit output versus 24? I'm curious if the slope of the filter needs to be increased with 24 bit output to prevent noise from being apparent in the extra resolution of the 24 bit word output.

Cheers

Kris

[Greggy]

Keep in mind you will have lower conversion latency at a higher sample rate. It's not just the extended frequency response that makes higher sample rates more desirable.

Is this because a buffer fills up with samples at a quicker rate than when running at a lower sample rate?

[Bassomatic]

24 (or 32 bit sometimes in Nuendo) and 44.1 here.