A) I'd go better if possible. Nothing - NOTHING will EVER be as important as your monitoring chain (period).

B) Most people can't hear the difference. If you were a bat or a dog and you were actually using headphones that went up to 50kHz (although as a human, you're limited to around 20kHz), then maybe.

But you'll find that survey after survey of full-time industry professionals, around 70-80% record at the target rate.

There are some advantages in processing at higher sample rates (and many plugs take care of that on their own). And some (few) converters tend to sound 'more open' at higher sample rates -- And undoubtedly, throwing calculations in 24 (or 32BFP, of course) bit drastically ups the resolution. But 16/44.1 is still a very reasonable delivery/listening format if it's done well.

(and of course, it'll only ever be done as well as the user's monitoring chain allows it to be)

This is an area where I need to improve as well. Here is the product and the thread where I think I found the best info for your budget:

http://www.amazon.com/gp/product/B001U92K6M/?tag=r06fa-20

{old-link-removed}

Good day.

sargentpilcher wrote: I am currently looking to upgrade the way I mix music. Currently, I have a fast track ultra and I'm just using a pair of 50$ monitor headphones. I have done test recordings of 16 bit, 44.1 Khz Vs. 24 bit 96 khz recordings, and I cannot hear a shred of difference. (I know my fast track ultra is capable of recording that bit rate). So I'm thinking it's the fact that I use crappy headphones. I would like to buy a pair of really nice monitors (Budget of 400$ max), but do you think I'll be able to hear much a difference in monitors that cheap? Or will I need to save some money and go all out on a pair so I can really start mixing better?

Thank you for any help!

A half decent set of speakers will be well worth the money spent, especially if your only using headphones now. You could probably get a pair of krk's or something comparable on Craigslist for a good price. Just make sure they are Fully working if you were going to buy them used.

As far as bit depth, I would suggest recording at 24 bit. I don't think there are any viable advantages to recording at 16 bit, but there are advantages to recording at 24 bit.

I do planning on getting a set of monitors, but my question is closer to, "Do I need a 2,000$ system to be able to hear the difference between the 2 different bit depths and sampling rates? " I'm sure any set of monitors would be better than just my headphones, but I want to start mixing at a high quality, and know the difference when doing so.

Thank you for all your replies. From what I can gather so far, it doesn't matter much as long as I'm recording and mixing in 24 bit, but I am definitely getting a pair of monitors.

I don't think you're understanding the physics here... If you record a 20kHz sine wave at 44.1kHz and another at 192kHz, they're going to be exactly the same. If you have a chain (and I mean CHAIN - Source, mic, preamp, processors, amplifier, monitors, ears, etc., etc., etc.) that can take in a signal above 22,000 cycles per second, then there you go. But high sample rates are driven mostly by manufacturers ("ours are better because the number is higher") rather than an actual need for those numbers.

Summing and processing aside (as some EQ's certainly sound better at higher rates, etc.). Calculative bit-depth aside (as the resolution during that summing is undoubtedly much higher in 24-bit at floating point depths).

But if you hear a difference between two recordings at different sample rates, then your converters are broken. And again - Most people - Most full time professionals - record at the target rate (44.1k for audio, 48k for video) - But basically all record in 24-bit resolution.

sargentpilcher, post: 297386 wrote: I do planning on getting a set of monitors, but my question is closer to, "Do I need a 2,000$ system to be able to hear the difference between the 2 different bit depths and sampling rates? " I'm sure any set of monitors would be better than just my headphones, but I want to start mixing at a high quality, and know the difference when doing so.

Thank you for all your replies. From what I can gather so far, it doesn't matter much as long as I'm recording and mixing in 24 bit, but I am definitely getting a pair of monitors.

Having the speakers or the environment that will help you to try to distinguish between something recorded at different sample rates does not really relate to being able to mix well. Being able to hear that difference should not be the goal. Balance and translation are a couple things to shoot for. The big picture, not the micro picture.

Are you more interested in 24 bit or 96kHz. As has already pointed out, 96kHz will probably not yield a significant difference. Although, I do record often at 88.2 or 176.4 and do hear a difference. 24 bit on the other hand, there should be an audible difference. However, all differences (bit or frequency) will be minute.

A $400 pair of speakers isn't going to get you very far. In fact, dare I say, a $400 pair of speakers in a bad room is no better, or even worse than your headphones.

Monitoring chain is more than just the speakers. I would say (purely off the wall) that speakers are 35% of the equation. A properly tuned room is 55%. D/A Converters, cabling and incidental electronics are the remaining 10%.

Do some reading in the acoustics forum, then listen to some speakers that are in your range and decide which works best for you. Treat your room (there are poor-man's treatments), add your monitors and hear a whole new world.

Cheers-
Jeremy

yeah just to +1 this forget 96kHz. Unless you have the most advanced signal chain known to man, any possible advantage will be destroyed when you go back to 44.1 for CD, and so you are better to make your choices at the target resolution. a very quick google will show you why to produce 16-bit audio you need your calculations to be higher, and also open up a never-ending world of calculations that will mean nothing unless your room is treated and you have a good D/A and monitor chain.

Just an alternative point of view

I'd like to offer an alternative viewpoint, and one emphasized by George Massenburg at a recent conference.

While many working professinals are recording at target rates, as professionals one can argue that we should be working at the highest resolution possible up to the point of delivery. I.e. for most people that would mean tracking, mixing and mastering at 24/96k, then downconverting to whatever the target rate/format happens to be.

There are other, valid reasons to use higher sample rates than simply audio information above 20kHz (that we can't hear anyway), f.x. gentler filtering much higher up in the frequency spectrum, and interaction between supersonic frequencies with ones we actually can hear, among others.

Additionally, for those (mostly the handfull of consumers labeled audiophiles) who have equipment to playback higher resolution material, there is a market (albeit small) for the higher quality as a delivery format itself. We recently completed a project tracked, mixed and mastered at 24/96, which will be delivered on vinyl, MP3 as well as high resolution files (24/96) for digital download.

What the others have said is correct. However as audio professionals I think it's up to us to set standards for quality (and also for archival purposes), regardless of whatever delivery format is in vogue today.

Cheers,
Thor

The point about our "not being able to hear anything above 20KHz" is that we don't spend all day listening to sinewaves. The energy in everyday transient sounds (bell chime, cymbal hit etc) has significant components above 20KHz, and we can sense the difference between this energy being present and it being removed, even if in steady-state terms, we can't "hear" it.

Several years ago when I got my second Alesis HD24XR, I did some experiments with splitting the output of my pre-amps into both recorders. I ran one at 44.1KHz and the other at 96KHz. I took the multi-track analog outputs from the recorders in turn and mixed on an analog desk (Midas), re-digitizing the stereo mix at 44.1 KHz. The mix done from the 96KHz recordings had more presence and air, and this was audible even after burning to CD at 44.1/16.

What this shows is that the typical processing involved in a mixdown (EQ, dynamics, effects etc) acquits itself better when it has the full range of the source spectrum to work with, even if the final delivery medium is restricted in both bit depth and sampling rate. For this reason at least, I advise people who want the ultimate quality to use higher rates for recording when there is likely to be significant transient energy in the acoustic source.

It is interesting that many critics rate highly most of the "direct-to-stereo" recordings, where analog mixing and other processes are done on the fly before the stereo mix is captured at 44.1 KHz for CD burning. This process is, in effect, a real-time manifestation of the high sampling rate recording and mixing before hitting the compromises of the delivery medium.

I agree with Boswell, unless you are at my end of the gear ladder, where transient response and "presence" are as abundant as good singers at an open mic night.

FWIW
To me this approach of 24/96k makes a lot of sense. It makes some logical sense because recording studios have always provided the highest quality recording equipment and processes to capture the essence of the music or sound being recorded. In fact if that logic makes sense and works then 32/196k should be the defacto professional standard! Not because somebody can possibly hear a difference but rather because that is the state of the art in terms of equipment available to capture sound in a well designed professional recording studio. Sure this may increase file sizes and create banks of HDD, but after all isn't this what real professional recording studios provide. Today, anyone can and will use a $1500 interface and a $1500 computer and call themselves a recording studio! What will be the point of anyone wanting or needing to go to a real recording studio unless the professional studio provides that same sort of equipment like the 24 track 2" tape machine did in the past? The only reason there were recording studios was because of the equipment, talent and the rooms. If professional studios want to rise above the fray, it would seem to me that the equipment, the sound, the talent, the digital resolution and computing power all needs to be at a level far above what Johnny can create in his bedroom! If not and everybody thinks or believes 24/44.1k is sufficient to achieve a given target product, then the professional studio recording industry will merely become mediocre at best and possibly fade out of existence...which is really what's happening.
What's needed is an industry of manufacturers who can come up with a professional level of equipment that far exceeds what's out there for everybody else!
This is all analogous to a recording studio tracking and mixing using a 24 track 2" tape machine and console then converting that to cassette or LP for delivery back in the day.
What some people propose is that you might as well have recorded and mastered on cassette to begin with because the end user has the same high end equipment the studio had to create the source material and has the same sound that comes off some high end monitor system and room.
From an industry standpoint that doesn't really make much sense....does it?

Well thank you all for your insights. Even though I can't really hear a difference, you've all inspired me to at least start recording and bouncing down at higher resolutions. If there's even just one person out there that will notice the difference, then it's worth it.

A couple of points to add. First, I think Boswell's type of test is the relevant one: Are there audible differences in the final product? I've done a similar test to Boswell's and did not detect the type of differences he talks about. But, I don't think my test was as well designed as his, and I would bet my ears aren't as good. Second, while storage is not a big problems, processing power is a problem for many home systems. If you are dealing with pop music with high track counts and having to deal with latency during overdubs a higher sampling rate can cause problems. Personally, when I'm recording an acoustic ensemble in one take with a few mics, I'll record at 88.2/24. For a multitrack pop or jazz recording I'll go with 44.1/24. If you don't have a top quality system, you can't do the same things as those who do.

Some exceptional $400 monitors are the Alesis M1 actives. I think for the money you can't get a better pair brand new. You can't really hear the difference in the sample rates IMHO. go higher sample rate pretty much just for the plug-ins.

SammyD, post: 299772 wrote: Some exceptional $400 monitors are the Alesis M1 actives. I think for the money you can't get a better pair brand new. You can't really hear the difference in the sample rates IMHO. go higher sample rate pretty much just for the plug-ins.

From what I understand, (At least in FL Studio, and I believe most DAW's), inside the DAW it's internal calculations are 32 bit. Not during playback, but when bouncing, and it then converts it to a lower grade format during the bounce.

That depends upon what you are bouncing to though normally it is 24 bit with a 32 bit float. This means that the bit rate changes as needed in order to conserve resources. And again, unless you have a high quality setup in a quality treated room with speakers that are capable of producing sound higher than 20k, it's quite a moot point.

Its pretty important that you record at 24 bits. It raises the head room so that you can get less noise under the mix. I've heard some big differences between 24 and 16 bit.

As far as the sample rate goes, if you have the hard drive space, you might as well go with 96khz, or even 88.2khz. It will give you undistorted frequency responce up to about 45khz, which is 25khz over the ceiling of what humans can hear, but it picks up the harmonics of what you are recording and it makes it feel more like an analog recording. Personally, i don't hear much of a difference in that.
Whatever you do, dont use 48Khz sample rate (unless working with video) I say this because when you are exporting to .wav (16bit 44.1Khz) it takes a really high quality converter to switch 48khz to 44.1khz. But its a lot easier and more bug free to conver 88.1 or 96 to 44.1

Basically if you can record higher quality, and have the space, then you should, but I wouldn't buy into 32bit or anything higher then 96khz.

Thats my 2 cents.

Your theory about converting 48k to 44.1k is no different than converting 96k to 44.1k. The easiest rule of thumb is to use multiples of the target medium. 44.1k becomes 88.2k becomes 176.4k. 48k becomes 96k becomes 192.

16 bit or 24 bit has nothing whatsoever to do with the quality of the recording. That is strictly a volume headroom component. Higher sampling rates (eg 88.2k) are useful for better resolution of plugins and mixing, but only if your monitoring room and signal chain are good. In an untreated room or with poor D/A conversion or poor speakers there is no advantage to the higher resolution sampling rates at all.

TheJackAttack, post: 344048 wrote: Your theory about converting 48k to 44.1k is no different than converting 96k to 44.1k. The easiest rule of thumb is to use multiples of the target medium. 44.1k becomes 88.2k becomes 176.4k. 48k becomes 96k becomes 192.

16 bit or 24 bit has nothing whatsoever to do with the quality of the recording. That is strictly a volume headroom component. Higher sampling rates (eg 88.2k) are useful for better resolution of plugins and mixing, but only if your monitoring room and signal chain are good. In an untreated room or with poor D/A conversion or poor speakers there is no advantage to the higher resolution sampling rates at all.

Bit depth effects head room and amplitude, yes. Higher bit depth=Lower relative Noise floor, which to me, sounds better. I've heard differences between 16 bit and 24 bit. I think 24 is important. Also bit depth is exponential so I think 24 bit is high enough, recording in 32bit is overkill.
As far as the quality of audio, to me, it doesn't seem like sample rate is quite as important. I still record in 96khz because I have the space, and why not, but it isn't a huge difference to record in 44.1 Granted, I have found that some DAW''s and plugins seem to run more smoothly with higher sample rates then others, it also seems to attribute to lower latency, as long as you don't bog down your cpu.
And the reason that i mentioned 48khz to 44.1 conversion is because I was in a situation with my recording professor at school where a studio who worked only in 48khz needed all of their tracks converted from 48 to 44.1 and it was a huge pain in the rear. It seems that because the sample rates are soo close the converter needs to work a lot harder. We were getting hardcore phasing issues because they were converting properly. It seems that the space between 96k to 44.1k is just easier for converters to handle. I'm not sure why, but I still don't recommend recording in 48k.

With respect, most of the classic CD's of the 80's and early 90's were mastered or recorded and mastered at 16 bit. The substantive benefit of 24 bit recording is headroom. Neither provides the sonic accuracy of DSD 1 bit performance. If your converters took issue with 48k -> 44.1k then they took the same issue with 96k -> 44.1. Probably the DAW you used had some older algorithms. In modern and up to date DAW algorithms this is pretty well worked out in either case but the best conversions will always be in multiples of the target. So once again, 88.2K is then better in regards to final destination of CD audio 44.1K/16 bit. If you were going to end up as a DVD or DVD audio then 96k would be preferable to render down to 48k.

Most programs are only 32 bit float which means they don't run at 32 bit constantly, only when needed. The saved audio is still in the 24 or 16 bit format as it was originally recorded.

Hi All,

First post here for a loooooooong time!

After the 96k rate became an established format, Glenn Meadows (mastering engineer and founder/owner of the SADiE and Mastering webboards) suggested that if you consider the improvement gained by 24-bit/96k over 16-bit/44.1k, 75% of that improvement would be the bits, 25% the sample rate, and it feels to me that the statement still essentially holds up.

In a conversation I had with Mike Kemp of Sintefex (Mike's a man with a brain the size of a planet who knows more than a thing or two about digital signal processing) he told me that for him the chief benefit of higher sample rates was in allowing more processing headroom and space for digital processes to operate. That also continues to hold good, as evidenced both by upsampling converters and plug-ins and by those mastering engineers who build up-/downsampling pre/post processing into their daily workflows.

As far as conversion of files from 48k to 44.1 there are huge problems, and I can explain them.

The issue is called, Quantization Errors.

Here is a situation. You record a song in 48k, that means you have 48,000 samples of each sound source per second, between each of those samples, there is no information of any kind. Now you want to convert it to 44.1k, or 44,100 samples per second. 44.1 and 48 have no relationship to each other so that means that many of the samples of the 44.1 tune would fall in the dead spots of the 48k tune. This causes your machine to have to do a bunch of Error correction that is far from perfect. This moving arround of your individual samples leads to QUANTIZATION ERRORS!!! When you sample in 88.2k there are no quantization error because it it directly related to 44.1, and when you convert from 96k, there are so many extra samples that the quantization error's are less evident.

Also, The Nyquist Theorem, States that you must sample at double the highest frequency you intend to capture. We mix down to 44.1k for cds because that gives us audio recreated up to 22.05 khz, which puts the aliasing distortion brick wall filter at that point. 48k only gives you and extra 2k, the differences are negligible. Its only used for video audio. I do suggest though that you actually record at 96k or 88.2k. Not because those extra several thousand hertz will make you're recording quality sound "more open, and warm," but because your plugins will function and sound much much better, because of the math involved.

So moral of the story:
Record at 96k, or 88.2k if you can. If you can't, don't fret, record at 44.1k. Just don't bother with 48k, i promise you it causes problems.
And also try to record at 24bit if you can.

NelsonReep, post: 354826 wrote: As far as conversion of files from 48k to 44.1 there are huge problems, and I can explain them.

The issue is called, Quantization Errors.

Here is a situation. You record a song in 48k, that means you have 48,000 samples of each sound source per second, between each of those samples, there is no information of any kind. Now you want to convert it to 44.1k, or 44,100 samples per second. 44.1 and 48 have no relationship to each other so that means that many of the samples of the 44.1 tune would fall in the dead spots of the 48k tune. This causes your machine to have to do a bunch of Error correction that is far from perfect. This moving arround of your individual samples leads to QUANTIZATION ERRORS!!! When you sample in 88.2k there are no quantization error because it it directly related to 44.1, and when you convert from 96k, there are so many extra samples that the quantization error's are less evident.

Also, The Nyquist Theorem, States that you must sample at double the highest frequency you intend to capture. We mix down to 44.1k for cds because that gives us audio recreated up to 22.05 khz, which puts the aliasing distortion brick wall filter at that point. 48k only gives you and extra 2k, the differences are negligible. Its only used for video audio. I do suggest though that you actually record at 96k or 88.2k. Not because those extra several thousand hertz will make you're recording quality sound "more open, and warm," but because your plugins will function and sound much much better, because of the math involved.

So moral of the story:
Record at 96k, or 88.2k if you can. If you can't, don't fret, record at 44.1k. Just don't bother with 48k, i promise you it causes problems.
And also try to record at 24bit if you can.

Re-sampling: your explanation of how it works is incorrect. 48KHz and 44.1KHz do have a relationship with one another, viz. 160:147, and commercial resampling algorithms make use of this.

There are many digital signal processing techniques that will generate a lower sampling-rate signal from a higher one, but none of the accepted methods uses the principles you describe. I would refer you to standard papers on the topic of resampling, e.g. [[url=http://[/URL]="https://ccrma.stanf…"]this[/]="https://ccrma.stanf…"]this[/] one from Stanford.

Sub-multiple rate resampling (e.g. 88.2K to 44.1K) can be performed by digital filtering followed by decimation, but although the mathematical processing is reduced, the results are no better than the general technique using a good algorithm.