Kara-Moon Forum

This one has me figuring for quiet some time: all of my hardware gear is set up to start working with 48Khz of sample rate.

A CD runs at 44.1khz, so I always have been working with this "non standard" sample rate.

Can someone explain me why the standard is not set to 44.1Khz, and which frequence rate is the better to work with? Is their uberhaupt an advantage using the one over the other?

I'm puzzled... ???


Moon

Another question is about bit depth, I usually work at 24 bit but I need conversion to 16 to write to Cd-r.
For the sample rate question, I don't know exactly why it's 48 Khz, I've always thought the higher the better.... ??? :-[ :-\

Moon & L.R.,

As I understand it,a higher sample & bit rate with to start out with(sound source,i.e.,instrument and/or recordering unit) is better to have-as long as the sample conversion process to a lower bit depth & sample rate is good,meaning that the sample converters in your gear is of high quality.

A good quality sample rate converter will of course,minimize any degradation in the conversion process,thus rendering your recordings with more detail & accuracy in the final result.

What really confuses me though,is that the sample rate of the Access Virus, is a whopping 192 kHz!!

It is possible though,to record 24 bit/96kHz quality onto a standard cd,using a studio grade,mastering cd recorder-but again,it requires a high quality sample rate conversion process to preserve the integrity of the original higher sample rate-but of course,the standard for most cd players is still 44.1 kHz & you would still have to convert from the master cd recorder to play it in a normal cd player.

I hope that sheds a little more light on your queries,but I myself cannot fully conceptualize the full scope of this process.

I am sure that Oren will be able to explain the whole sample/bit depth concept much better than I can,as he is a much more astute musician when it comes to technical knowledge.

Elwood

I don't know how much help this is, but, yes, a standard CD is 44.1 and 16 bit. So, eventually, if its going on a CD, it will end up at that standard.

Not sure why some equipment is set at 48. Some old Soundblaster cards were set at 48 only. Would cause occasional problems when shared and used in a 44.1 projects.

I've heard that bit depth is more important relative to sound. I think the current standard for recording is at 44.1 and 24 bit. Of course, the bit depth will have to be dithered down to 16 bit for the general CD consumer. 

Bill (deaf dunderkwac) has excellent knowledge in this area, so perhaps he'll chime in. 

I'm by no way an expert on this but have done a little reading.

44.1kHz for CDs was a bodged job. The original digital recorders were converted VIDEO recorders, and the frequency is a match for both the old US and UK line/dot-rate standards.

DVDs are apparently 48kHz which is much more sane, so you should give this some thought if you are recording stuff of great cultural worth (I wish!).

192kHz is actually a good figure to master a recording at. It divides exactly down for 48kHz, and is close enough to 44.1kHz for good resampling kit to make a sensible attempt to iron out the errors.

I can tell you this- the sample rate (44.1kHz, 48,etc.) determines the highest actual frequency that will get handled by the sample, and that will be approx. (very close) to one half of that rate.
 So a 44.1 rate can only handle signals up to 22kHz.  A 48kHz rate can handle signals with a frequency range up to 24kHz.  Etc.

 These samples occur at a speed determined by the rate (48kHz is faster than 44 kHz) therefore each sample it grabs is apt to be a more accurate representation, but mainly occuring in the frequencies we can't hear anyway, and there will be more of them. Like using a ruler with cm markings compared to one with mm markings.

The bit depth is actually a determiner of how 'wide' the each sample will be. I like to think of it as determining how big of a chunk each 'bite' will be. So the rate is how many (or often) each sample is taken, and the bit depth is how big of a chunk is taken. Bit depth determines the dynamic range. 24bit will be able to hold more dynamic information than 16 bit.

 These two factors can add up to bigger filesizes and more work for the system, and there is always a point of diminishing returns, where the tradeoff for filesize does not justify the quality increase.

 Many good reads on this stuff using Google.

As mentioned, the key is how it all sounds once it's in the 16/44 format, as that is the common CD format being used by everyone.  The 24 bit DVD-audio format is starting to show some promise.

And folderol is half-right about the 44.1 format. It was video speed-based.  It does work out very well for video, and just happens to be capable of incorporating the frequency range most of us can hear (the famous 20Hz-20kHz), but I think the format has served us very well for many years. The cassette tape, 8-track tapes,etc. didn't last as long as a format or serve us so well.  Now that everybody has a system capable of 24-bit recording and playback lots of people say '16/44 was a bad choice' but when I started computer recording 18 years ago it was rare to be able to handle 16/44 format.  It was like 24/192 of today. I had to buy a special soundcard that handled the conversions because those old original 386 processors certainly weren't going to!  The Roland RAP-10 audio card and software was amazing in it's day.  I had an old pre-pentium machine running Win95 that was capable of multitracking CD quality audio!  The harddrive was the big stumbling block- I started with a 100 meg drive, which could only hold a few tracks of audio...
so the format was great for the times.
 We are due for an update to the 'standard', and to me the 24/96 format seems future proof enough, since it can handle a wider range than we can hear..although 24/48 would seem more than adequate for now.
 Actually for recording tracks, most of the pro guys I talk to have resorted to 24/48 because after that you just add filesize and not much in the way of quality..

My system can do 24/96 multitracking,and my CD/DVD player can do 24/96 too but I don't see any sense in it, harder on the system, the hard drize especially. I'm just lately going to the 24/48 standard myself..
 
Here's a fun read on it; some people are very attached to their feelings on this, but Lavry seems to have studied it pretty thoroughly..
http://recforums.prosoundweb.com/index.php/mv/msg/4097/0/0/0/

I won't add anything to the theory since everything has been said allready.
Except... i've read that if you record music where the final format is a CD (44.1 kHz) you better use a samplerate which is a multiplication of 44.1. Some soundcards/systems can handle a samplerate of 88.2 rather then 96 kHz.
Reason for this is supposed to be the calculation errors which you get when you sample down the final product to the CD which is still 44.1 kHz.

Personaly I think that those higher samplerates are 'most' a hype thing put on the market by the companies that sell hardware. I won't deny that it is higher quality, but I could discuss the fact if you actually hear the difference.
A blind test could show this.
Humans can only hear as much as they can.... the proof : take the recordings of the tracks of the In a spin CD, for those who have access to those, download the 3 formats, shuffle and play them in a blind test and try to make out which one is in which format. I've done this, I can take out the wav file of our proper song but can't separate the MP3 from the OGG file.
For another song, I can't do this.

I still work at 44.1 kHz:16 bit, it has one advantage. The recording will sound exactly as I hear it when I record it, because there is no conversion.

k

I've worked at 24 bits, 48kHz for a few years now (previously I worked at 16 bits, 44.1 kHz). As far as I can tell, the extra few Hz do offer more transparency, especially at higher frequencies. The extra bits help mostly in dynamics as you have more leeway and run less risk of clipping when summing signals. In that respect I've been thinking of going to 32 which has been suggested in some magazines.

Having said all that - I guess you can make good or bad recordings and mixes in whatever format you choose so I won't go as far as saying that you must always go higher in quality. For example, when mixing rock or pop you don't need have that much difference in dynamics (everything is compressed and limited a lot anyway) so 16 bits could be fine. Alternatively, the wide range of dynamics make acoustic instruments prime examples for the higher sample/bit rates, especially in recording.

This is a very good thread, btw!

Well, I am glad I am not the only one.

I would not argue that there is no quality difference in theory,
but as a matter of practical fact, I cannot hear it.

I have yet to meet the person who actually can hear 24khz.

Wyatt

@ Bob: Thanks for a very clear explanation of samplerate/bit depth.

A quote from BobKatz in the prorec forums-

"There is no A/D converter in the world which actually outputs floating point as a format, in fact, there is no reason to do so, as the entire outside world can fit comfortably within about 20 bits. And most A/Ds only have an SNR approaching 20 bits, so 24 bits fixed is more than enough.

There is NO BENEFIT WHATSOEVER in recording to 32 bit float format and at least one drawback (additional storage space). Any DAW that requires you to do so in order to perform its operations more efficiently is just doing it wrong."

just sayin'..he knows his stuff..and i've read that repeatedly by others also..

Then of course comes the problem that most plugins,convertors, and codecs are not equipped for 32bit use.
 One person on InaSpin used 32bit on the original wavefile, and I had to search out a codec to try to find a good conversion to a format my mastering plugins would work in so I could master it. When I originally opened the file my system just went 'huh?" ???

 Anyway, 24/48 to 24/96 seems to be plenty, and if you have the 24/88 option some claim the math is smoother when the dither down occurs.

Wyatt- about half way down the first page of the article I posted above a 'Robert Randolph' claims to have been involved in testing where things above the hearing range affected the sound.. not sure I can agree with him.. but thought you might be interested.  That thread wanders thru a lot of interesting things regarding rates/resolutions.

That's actually a good point. Now that you mention it, I believe that Cubase internally mixes at 32 bits and I think Sonar even does at 64 bits internally.

But like I said, there's a point where the audible advantages - whether theoretical, practical or even hearable only by a type of hearing trained monks specially trained in Tibet to hear the fall of a mosquito on the other side of the world - don't weigh up against all the cost in terms of disk space, time and trouble. Though I'm still not sure 24/48kHz is the end all in that regard; I think audio will end up one step higher as a consumer format in a few years (and consequently in our DAWS) to REALLY make sure all the audiophiles in the world are satisfied. ;)

Then of course we will get all those retro VSTi's which sound like the VSTi's we have today to get that classic 24 bit sound... :D

This thread has really progressed in a really pragmatic way & there is a plethora of very useful information,of which reminds me of some of the technical advice that I recieved from the user forum & sales engineers from Sweetwater.

I do agree with Kara that there is some techno-hype out there,especially with regard to bit depth and I think a prime example of this, is the differences in sound quality between the 24 bit audio recorder in the Alesis Fusion,vs. the 16 bit audio recorder in the Korg Oasys.

Clearly,the Oasys has much more detailed audio clarity than the Fusion,due to the Oasys's higher quality components overall.

In reference to specifications such as % of THD & frequency ranges,there's plenty of hype out there on paper,as some of the spec-sheets out there boast some rather drastic levels,as from one company to another,measurements are based on different decibal levels.

My Adam A7 monitors's specifications read from 45Hz to 37kHz-which seems ridiculous on paper,but the way it was explained to me,is that despite the fact the the human ear cannot hear past 20kHz,the broad frequencey range of the Adams give you more overall clarity & detail within the range that is audible to the human ear.

I used to own the Samson 50a's & I loved them & I still think they are fantastic monitors for thier price & the frequency range read something like 70Hz to 20kHz-which seems much more realisitic on paper,but there is an enormous difference in overall stereo imaging,bass articulation & more detailed high end that I can actually hear.
So that being said,at least some of the specs out there read true.

What I find to be really curious though,is the unbelievably high end frequency range of the Tannoy monitors which read 50kHz!!

If there is any sort of difference between my Adams & those Tannoys,I don't imagine I would be able to hear the difference and at Tannoy's pricing,I don't care to find out.

Before I really looked into the Alesis Ion,I was not expecting the Ion to have the same sonic clarity as my Korg Triton,but to me,it sounds every bit as nice as my Korg,so a person cannot really make sound judgements based on pricing & specs alone obviously,as there are so many variables to consider.

Elwood

One other thing I forgot to mention,is that I was really curious about bit depth,with regard to DSP effects processors that are built into analog mixers.

I was considering the Mackie 6 channel DFX mixer,as it seemed to be a rather clean mixer overall,with 32 bit effects.

32 bit effects seemed like hype to me,especially since my sales engineer told me that stand-alone,rackmount effects processors are cleaner sounding than any effects processor you would find in a mixer.

I already have plenty of effects to work with in terms of recording with my Ion onto my Zoom recorder,but since I am also planning on getting the Nord Lead 2X,I would like to have the ability to have plenty of effects at my disposal,without having to go through my Zoom recorder,if I am just playing around with my Ion & my future Nord in general-as supposed to recording.

After doing more research,I discovered that the Samson MDR 1248 12 channel mixer
with 256 preset effects,would be an ideal choice for me,as it would nicely merge together,my two keyboards,since my Zoom recorder only has one input.

I am really growing tired of unplugging one keyboard for another,especially in the event that I want to play both at the same time & record! >:(

Considering Samson's features for the price & a seemingly reasonable level of quality overall,I cannot think of a better mixing solution,than this mixer. 8)

Elwood

Just an addittional note from me.

In my yoof (about 18-24 actually) I and some friends, did a lot of practical experimenting with the 'new' semiconductor devices of the day. They were called SILICON transistors :D

Our experimenting actually had to be practical 'cos none of us had the money to buy expensive test gear (A 'cheap' 'scope was about 10x our anual earnings).

The first thing we noticed was that sensitivity to distrotion was highly depended on both the type of distortion and the frequency of the signal being distorted. Simple harmonic distortion comprising mostly 2nd & 3rd harmonics remained virtually undetectable up about 2% for middle and upper frequencies, and about 0.5% if the signal freqency was around 100-200Hz. Both of these figures increased for complex waveforms, with the exception that signals with a  lot of HF (cymballs, etc.) started to sound 'muddy' very quickly.

Step-wise distortion, aka crossover distortion, on a single tone we could detect very quickly, and if there were two or more tones, we could detect it at (to us) unmeasurably low levels because of the intermodulation signals it created.

These days I don't have the time, facilites (or hearing) to do such layman's tests on digital systems, but I suspect the stepwise nature of the signal would on the face of it produce noticable results at low bit densities (16 0r less), but that this would be partly mitigated by the high 'chop' frequency.

What is not at all clear though is how intermodulation would be affected (step-wise stuff always produced some of this). I have heard it said that having a wide signal input bandwidth can actually make this worse. Only, I think they call it 'aliasing' these days.

Something along the lines... if you have an 19kHz and 18kHz signal any harmonics would be way out of range, as would the upper alias of 19+18. However the lower one, 19-18 = 1kHz, or slap bang in your most sensitive range!

Well this is all a bit of a ramble, and I don't have any hard and fast statistics. It would be nice to know ;D

There a lot of info in this thread and this is a very interesting topic.

But, let's come back to this question: When the music is meant to put on a cd, what is the sense of working with 48Kz, opposed to 44.1Kz?
I do believe it's better to work with 24 bitdepth, leaving more material to work with for the mastering. However, I still fail to see the point to work with 48Kz, since you will need to convert this afterwords, affecting the sound quality...

The gain of quality off 96Khz (88.2kz) or even 192Khz is marginaly. Yes, there is a difference, you can't argue about that. But how many people have a stereo where you will actually will hear the difference? And how many of those people have actually a good acoustical room to listen to this high level of audio? Only a very few purist with lot's of cash...

Moon

Moon,

I don't see why 24 bit would not be better to work with in general,provided that you have good A/D & sample rate converters,but my whole point was,is that the Fusion was poorly designed with poor A/D converters,so how is it that the Fusion could sound better than the Oasys-just solely because of the 24 bit of the Fusion?

As I see it,there are no compromises in build & material quality in the Oasys whatsoever and Korg spared no expense in the manufuacturing of the keyboard-hence the heavy price tag.

If the Oasys had a 24 bit audio recorder instead,then I am sure that this bit rate would improve things even further,coupled with the high quality A/D conversion.

With all these aspects covered in this thread-it raises yet,even another interesting question in my mind...which is that if there is little or no benefit to using 48kHz & then converting to 44.1,then what is the advantage to using 24 bit,if it's going to be converted to 16 bit anyway?

I'm really getting lost being so immersed in this discussion...somebody help!! ??? :-\ ;D

Elwood

There is really not much difference in those two sample rates..
 maybe a tiny bit of sonic headroom... but I can't give a definitive answer on what you will hear in 48 that you won't hear in 44.1
 When it comes to a synth-I guess what would matter more is the convertor being used..maybe some 48 convertors are a bit better than 44.1, so the actual quality of the transfer from audio to digital (or digital to audio) is better (?)
Maybe 48 convertors that sound (technically) better can be had for a better price..
Maybe the architecture of the way samples are handled inside that particular synth works better with the 48 convertors..

 Elwood- the change from 16bit to 24bit will be more noticable than the change from 44 to 48..

 From 16 to 24bits is only an 8bit increase, but that's half of the original 16 bit!

Maths:
16 bit is a value between 0 and 65.536
24 bit is a value between 0 and 16.777.216

Meaning: you don't get half us much more detail, but you get 256 times more detail. So, yes, there is a very big difference between 16 bit and 24 bit !

24 bit is giving you much more headroom to work with, which is good to keep the S/N raitio to a minimum and which offer plenty of room if your music has silent and load parts.

Moon

Yeah, it's funny how exponentially, what sounds like a little is a lot.
Thanks for giving the actual numbers..puts it in better perspective.

I've heard this same reason for using 24 bit, on several occasions.

some one say my name lol?

quick rundown...
44.1 legacy standard for CD chosen at the time because it was the best compromise of the available technology.

...along comes digital audio-video recorders and a standard had to be chosen and it's hard to divide 44.1 into 30 fps (or, if you will, 29.97 frames per second (fps) rounded to 30 for teh easy math types such as meself) so somebody chose 48.

by the time rofessional digital video came along technology had surpassed the old CD standard and the techno's finally figured out that 16 bit was just too hard (cost-wise) to filter so they went for 24 bits. Important to remember that the bits are the number of slices taken each sample. 8-bits would have 256 slices and you can look up the rest <grin>. The filters have to smooth the transition between slices and during the record process, not allow any energy above 1/2 the sample-rate (otherwise aliasing would happen) The higher the sample rate, the easier it is to filter out the bad stuff) (the higher the bit-depth the easier it is to smooth the steps)

Many audio purists thought 44.1 @ 16-bit was <horribly>  inadequate for sampling (recording) as all sorts of distortions were added to the signal so there was alot of experimentation done in the 90's.
recording a sound at 192kHz @24-bit is like passing the signal through a short piece of wire... there was no deterioration at all!
But it was later shown that most of the deterioration of the signal was happening in the filters (used to smooth the 'steps' of the slices
-Better filters were developed...
If you do this for a living I would hazard that recording at 96 or 192 kHz@ 24 bits would be 'the thing to do' but any bit rate 20 or above is easier on the golden-eared peoples of the world.

You can always down-sample (convert from 192 to 96 to 48) with impunity but if you up-sample you're not gaining anything. You cannot create bits that aren't there.

real-time Sample-rate converters  (SRC) convert the digital signal to <gasp> analog then re-sample at the new rate (actually, they sound pretty good nowadays)


The above information is really only for those who sample (record) real instruments
I use vst's which don't have the same issues (instead they have other issues and I don't even wanna discuss the vst's based on samples :D )
 hope this helps a little.

D.D.,

Your very comprehensive,yet very clear & easy to understand info & explanation was a tremendous help to me.

Thanks for sharing your technical knowledge. 8)

Elwood

You're welcome

a last gasp at the subject.

does anyone still make cd's? Hopefully your sound interface can work at multiples o 44.1 (88.2 et al) If not there are not-so-realtime src's that do an excellent job of conversion.
My host does an ok job of mixing SR (sample rate) and media types (mp3 wav and ogg) but in the recent past most hosts 'could' mix and match but at varying degrees of success.

It's usually better to convert all the tracks to the same SR and media (wav preferred for teh windz users) as that takes quite a load of of your computer's processor(s) so you can ad MOR EFFECTS.

Thanks, Bill!