This technical blab la la is great BUT becomes quite stupid if you consider that an average artist (player) don’t care about this at all. Fact is that even cheap sound card on motherboard can produce pro quality sound. Only what is left is sound engineer which can damage this sound. What I was trying to say is that only human ear can answer on this question. If you, like a sound engineer, with same hardware can’t hear difference way you are asking this question on the first place?
I think that we are little outside off the road. GOOD QUALITY SOUND is the sound which is nearest to the sound which was imagined on weary beginning of project. If sound engineer AND author of the song are satisfied with sound quality who cares about sampling rate at all. If you like sound engineer can’t find a job then definitely you have to change something.
Once again -> do complete project in 44.1 and then start again in 96 KHz (today standard for recording) and then decide what your way is. Attention -> don’t try to repeat exactly the same settings for fx and plugs – use your ears and definitely you will see difference.

 

First facts then opinions? In that order?

 

Thanks Muze!

 

Just a question, guys:

If it is much better to downsample from 88,2 to 44,1 kHz (which sounds very logical), why is 96 kHz the standard in pro studios nowadays, and not 88,2 kHz?

Or is it not true, that 96 kHz ist the standard?
Fact is, more soundcards support 96 kHz than 88,2 kHz (mine too, btw).

 

Perplexed eh? That's cool, I like to puzzle people. But seriously, how do you determine who is a primary or secondary source or who has or has no experience? Do you know Felino personally? If not you are assuming, methinks.

That's cute
However, I did say that some of the stuff from that Reason forum was poorly explained and some even garbage so I am somewhat at a loss seeing your need to repeat it in such manner as if I wrote it. I don't discount all the information I read just because some of it contains rubbish. If you dismiss something on that basis, you ought to get out of Detroit right now because half of it is a ruin so the other half must be too...

True, and twice the number of unwanted digital noise that now needs to be filtered out too. When you downsample back to 44.1 KHz you lose that higher resolution again unless you have a really sooperdooper resampling filter. I am not sure which of the two posts you say is wrong but the author of the second one I posted, Martin Russ, published his book three times now without anyone protesting. So, I guess you must be referring to the smaller post.

That is true too, but remember that it incorporates an absolute humongous amount of redundancy, especially at the lower end of the spectrum. 96 KHz only gives you 4 samples at 20KHz as opposed to 2 samples at 48Khz. Because of the very high frequency no human is going to hear that difference. At the other end, in the bass range, you are going to have a substantial increase of samples that are not going to make much of a difference either. an acoustic Piano A at 110 Hz is not going to sound any different played back at 48 KHz than it would if it were sampled at 96 KHz. The only area of interest would be the range between 2000 Hz and 8000 Hz. Beyond those ranges no one is going to hear the difference between sample playback rates of 48 and 96 KHz. And at any rate, by the time the music is commercially released it is downsampled to 44.1 KHz anyway.

agreed! And resampling ALWAYS degrades the signal.

That would assume that you have an "intelligent" encoder. Somehow I don't think those are used outside of science and military laboratories

yep.

Okay, here you lose me... WAV/AIFF are the two native windows/Mac formats that write raw audio data to a hard drive, so that would be the original unaltered sound on a computer which is ALWAYS digital. I don't really know what you mean by "natural." The only other way to store sounds is on tape where it would be analog and which would be a major degradation over the digital domain since tape adds a ton of noise. The headroom of tapes is quite low, typically around 70 to 75 dB between the noise floor and tape saturation. That is comparable to about 11 to 12 bits of resolution in today's digital measurements. I am sure, tape is not what you had in mind, so what did you mean?

My personal opinion about recording sample rates: use high rates only for Classical Music in a near-perfect recording setting and meant to be mastered to CDs. Anything else including electronic music and dance music 44.1 KHz. Because of the high compression used there is really nothing in modern music that warrants a higher sampling rate. Don't even think of using 48 KHz if it isn't for Video or film.

 

Very interesting reading.. thank you.

Did you know that blind listening tests proved that 9 people out of 10 could not differ a WAV music file with the same in MP3 (320kps)? And no-one can tell the difference between a music file at 44.100 and the same file at 96.000? And also that there is no playback system yet capable of reproducing the "benefits" of 96.000kz recordings.
There is a proven gain in quality going from 16bit recordings to 24bit recordings (lower floor noise major aspect) but no tangeble difference going from 24bit to 32bit recordings.

What we could all do with is a new type of CD (re-write the Red Book please...) capable of playing back 24bit recordings, just like DVD.

 

Jayar,

If everything is recorded digital at 44.1 and at 88.2 and processed/mixed/mastered simultaniously there is a difference and people will confirm the difference. But will disagree on what is better.

A Moog bass sampled at 32khz versus 44.1. may sound "warmer" . When essentially everything above 15KHZ is removed (at 32khz samplerate) from the source.

Because of this logic best is to stay true at the intentions of the artist....

 

If phase is something, best present at higher samplerate. What is the place of phase in a stereo recording.... Well, vibrance. Wich can be gotten from M-S recordings. Location detection is not needed. Locationdetection becomes a part of surround recording. I believe the step from 44.1 to 48khz is a big one soundwize in stereo. 96khz is great for surround. Wich essentially is for SOUND TO MOVIE. As SACD and DVD-A seem dead formats...

This is my logic. Yours may differ. In the end. The logic from the artist is what count for me... (unless I can get away with mine).

 

I also can't tell the difference between a WAV and a 320 kb/s MP3 (yes I have the guts to admit it), at least not on the monitors and headphones I possess. And I know that I have very good ears compared to other people (not saying that there aren't surely lot of people with better ears than mine).

@ Caithleann

May I ask, why you drew this conclusion?
If you claim that it makes no difference if we use 44,1 or 96 kHz, why use high rates on Classical Music?
That somehow implies that there IS a difference, which depends on the musical source.

Could somebody relate to my question from my last post, please?

 

I am also no longer able to hear much beyond 17 KHz as age is catching up up with me and temporary threshold shift has become more permanent these days thanks to the loudness of modern Dance music.

The Redbook CD standard can not be rewritten, that is why there are different colors in these standards, look up "Rainbow books" The Redbook standard was designed specifically for the audio CD which could hold exactly 1 hour of 16-bit music which would be 600 MB in size. This was based on the density of the pits (logic zeroes) that could be etched on the surface of a CD medium. Things have improved in a big way with the arrival of DVDs and Blue Ray but the CD remains fixed at that density, for legacy compatibility.

 

Interesting thread!

 

I found something to add to John C's earlier statement of

It is a good assumption. If the converter quantizes arbitrary values to fill up the "vacuum" between two adjacent values to accommodate the up-sampling process chances are that a good bit of aliasing could be introduced where there wasn't any before.
When down-sampling a significantly oversampled stream that picture changes somewhat when the sample rate is reduced by a process called decimation. The Low-pass filter increases the word length of the samples by an accurate calculation of each sample based on the values of surrounding samples. In this case there IS existing data that gets averaged before it is replaced by a single quantized value as the sample rate is reduced.

Needless to say that this is a very time consuming process and often not worth the effort if the gain in fidelity is low to negligible.

 

That I knew already, but thanks for spelling it out for all.

But the technology is there and I don't believe for a moment that a "new" type of CD, compatible with the old standard, could not see the light of day.
It could be done but it is not economically profitable. That's all.

 

It is quite interesting to see you use WAV and MP3 in the same breath. MP3 and most other audio compression protocols
are "lossy," meaning they lose something while compressing audio. That something is the very stuff that gave rise to this discussion in the first place, namely additional audio data that was included in the final recording after have been sampled at a higher frequency than the usual 44.1 KHz standard. The only media that would be able to play the files with increased fidelity are WAV, AIFF and Sound Designer II audio files because they are uncompressed audio formats. Using the MP3 protocol to save files negates the added fidelity because it removes psychoacoustical data. Psychoacoustical means 'perceived audio.' Much of the data in an audio stream, while present physically, cannot be heard (perceived) by our ears and auditory system and is therefore deemed expendable by the MP3 algorithm and the higher the compression ratio, the more data is removed. The so-called added benefits of higher sampling rates are usually the first casualties. The entire debate about higher sampling rates is moot if any of their proponents use any other audio format than WAV, or AIFF to save data. (Sound Designer II is quite obsolete, it is the predecessor to pro Tools.)

What conclusion, Gulliver? You don't refer to it and there are a few

You're right, there is a major difference between classical and all other forms of music. And while I am certain you are familiar with this, I add the following for new audio users.
Besides the frequency range of 20 - 20,000 Hz that a human can hear there is also a dynamic range that varies between absolute silence (minimum audio threshold)at 0 dB SPL (Sound Pressure Level) and the pain threshold of between 130 - 140 dB SPL. That is a very large range. The capabilities of recording/reproducing gear from microphones to loudspeakers and everything in between limit the usable dynamic range for recording to about 115 dB.
A bit depth of 16 bits yields a maximum range of about 92 dB so, in order to capture the entire dynamic range the next level of bit depth must be utilized, 24-bits. 24-bit is usually associated with 96 KHz so this explains the "discrepancy."

Absolutely! You asked

96 KHz is not the standard, but some studios use it as their internal standard, if they produce CDs for the consumer market they will have to record at 44.1 KHz.

There are two standard sampling frequencies between 40 and 50 kHz: the compact disc (CD) rate of 44.1 kHz and the so-called ‘professional’ rate of 48 kHz. These are both allowed in the original AES5 standard of 1984, which sets down preferred sampling frequencies for digital audio equipment.

The 48 kHz rate was originally included because it provided tolerance for downward varispeed in digital tape recorders. When many digital recorders are varispeeded, their sampling rate changes proportionately and causes a shift in the audio baseband. If the sampling rate is reduced too much, aliased components will become audible. Most professional DAT recorders allowed for only around ± 12.5 per cent of varispeed for this reason.

The 44.1 kHz frequency had been established earlier on for the consumer compact disc and is very widely used in the industry. In fact it became the sampling rate of choice for most professional studios.

Hope this answers your question

 

Great stuff, Caithleann,, great stuff indeed. Very informative yet to the point. Thank you.

 

I understand your point, but I never said that one should record higher than 44,1 kHz.
I also don't exclude the possibility that I would hear a difference on top-level monitors, just saying that I can't hear a difference on the monitors I have right now. But I also bet that most of the people who always claim that MP3 is shit, never made a blind test between mp3 and WAV. Or they are referring to experiences made with 128 kb/s MP3s.

Sorry, I was talking about this conclusion or opinion of yours:
"My personal opinion about recording sample rates: use high rates only for Classical Music in a near-perfect recording setting and meant to be mastered to CDs. Anything else including electronic music and dance music 44.1 KHz."

But you explained it later, talking about the dynamic range:

Well, this is a new information for me, that 24-bit is associated with 96 KHz. I thought so far, that the bitdepth and the sample rate has nothing to do with each other.
Does that mean that the full benefits of 24-bit can only be used when recording at higher sample rates?

Thank you very much
But unfortunately, not completely. If people talk about higher sample rates, they usually refer to 96 KHz, and very seldom to 88,2 KHz (at least that's what I noticed). Also HQ vinyl rips are made usually with 24/96; and lot of soundcards support 96, but not 88,2 KHz.
But why, if very likely at some stage you have to downsample to 44,1 KHz, and for this reason 88,2 KHz would be the better option?

 

I realized it much later when I saw the 24/96 or 24/192 mentioned in relation to modern day recorders but never a reference to 24/44 or 24/48 as standards. That is when I started digging deeper.
The Redbook standard is based on the 16-bit (92 dB SNR) resolution and a 44.1 KHz sample rate criteria. With those parameters you can create 1 hour of music or about 600 MB on a CD and that was the standard in the 1980s and 90s. Although 24/48 is technically speaking possible I suppose, it cannot be used to write CDs. Adding just 1 bit to the bit depth to go from 92 dB to 98 dB would significantly increase the size of the recording and exceed the CD's storage capacity and break the standard.

bit depth and sample rate both affect the size of the audio file.

The reason they use that high bitdepth and sample rate is to prevent losing any audio quality.
Down-sampling from 88.2 makes mathematically much more sense than down-sampling from 96 KHz. Sampling and quantizing filters have much less work to do and the error margin would be a lot narrower. But this was true 20 years ago. Nowadays these filters are so sharp and effective that the frequency increase hardly matters anymore. The newer and better filters have brought out something that used to be drowned out below the noise floor: jitter. Today's big problem is jitter which is a variance in the clock speed of the sampling device. If you record a clean sine wave with a sampling device that has a jittery clock and then play back the sample with a clean clock, the result is a dirty sine wave. Here a sampling frequency of 96 KHz can bring distortion into the picture and this jitter becomes even more of a problem the higher the sampling frequency goes. So, jitter is a new phenomenon where an increase in the sampling frequency can produce an increase in distortion.

 

Hello,

i tried something, i opened omnisphere in sonar x1 and played some notes one time with 44.1 khz and one time with 96 khz and i can hear a difference, 96 khz is a clear sound while the 44.1 is duller. i played the same midi file for my girlfriend and asked her which sounds better she is a musician plays instruments so she has a good hearing, but i dont told her which is 44,1 khz and which is 96khz but she could tell me that the 96 khz sounds better not harsh as the 44,1 khz.

So for me the question is answered now.

cheers

 

Amen.

Caithleann is NOT a woman, simply because no woman can be so spiritual.

 

I'm sorry but this seems like absolute nonsense to me.
Especially if you play a patch in Omnisphere which uses samples... if the samples were recorded in 44,1 KHz (which I believe they were), they will not sound better if your DAW is set to 96 KHz.

But everything is possible, so why don't you make two audio-mixdowns, one with 44,1, and one with 96 KHz, upload it for us to hear, and let us decide by ourselves?