Recording at 96/24 when other elements are 44.1/16??

  • Thread starter Thread starter BeyondMusic
  • Start date Start date
B

BeyondMusic

New member
I have some soundfonts and various samples in a Cakewalk Sonar song but I want to record the whole mix together in a separate track at 96khz/24-bit, but it keeps defaulting to 44.1 (the quality of the samples and soundfonts).

I have a Delta 66 soundcard.

How do I resolve this? How do most people record their final mix in best quality possible?

Thanks!
Peter (VoicOfReason)
 
The sampling rate and bit depth have to be the same with SONAR. You can change the bit depth, but not the sampling rate. You would have to export the audio to an application like Sound Forge or Cool Edit, change it there, and then pull it into a new SONAR project file.
 
Peter,

> but it keeps defaulting to 44.1 (the quality of the samples and soundfonts). <

Even if there was a meaningful difference between the quality of 44/16 and 96/24 - which I doubt - surely there will be no improvement when the original material is 44/16. Since 96/24 uses three times more disk space than 44/16, why do you even want to do this?

--Ethan
 
Huh?

Then why is there a huge market for 96/24 soundcards at all if everyone is in the end going to mix down to 44.1/16 CD format or even MP3 format?? Of course there is a difference. You record tracks at high rate, add EQ and effects at that high rate, then convert to lower rate.
 
Re: Huh?

Peter,

> Then why is there a huge market for 96/24 soundcards at all if everyone is in the end going to mix down to 44.1/16 CD format or even MP3 format?? <

Excellent question. Much of the push toward higher sample rates and bit resolution comes from the companies that sell the hardware, and the magazines that sell ads to those companies.

> Of course there is a difference. You record tracks at high rate, add EQ and effects at that high rate, then convert to lower rate. <

Have you ever made a direct comparison yourself?

--Ethan
 
Comparison

Yes, I have made the comparison. That's exactly why I bought a 24/96 m-audio delta 66, because there is a BIG audible difference between 44.1 and 96. Even on my little desktop speakers I notice a difference, but especially on studio monitors.
 
Re: Re: Huh?

Ethan Winer said:
Excellent question. Much of the push toward higher sample rates and bit resolution comes from the companies that sell the hardware, and the magazines that sell ads to those companies.
Click to expand...

he speaks the truth, as someone involved in PC hardware and some hardware review sites, it's all about pushing the product and little about whether it's truly useful or innovative.

24 bit is slightly better, but most people's ears can't tell the difference. I personally find 16bit recording to be more than adequate.
 
ok

Probably so. I have sensitive ears and can tell the difference. I can actually hear the silent alarm systems when I walk through malls though they beyond the normal range of hearing, so that probable explains why it is a big difference to my ears.
 
Re: ok

B,

> I can actually hear the silent alarm systems when I walk through malls <

Your ability to hear very high frequencies has nothing to do with bit depth. And I'm skeptical of anyone being able to hear the difference between 96 and 44.1 sample rates. Unless your speakers work past 22 KHz. (most don't) and your ears can hear past that too.

More to the point, your original question related to bouncing 44.1 source material, where nothing will be gained by using a higher destination sample rate!

I'm curious about the details of the test that convinced you 96 offers an audible improvment over 44.1. What did you record? Was it the exact same source material recorded simultaneously at 44.1 and 96 on two separate systems? What was the source?

--Ethan
 
skeptical?

Well I can see that you have a strong opinion about this. I'm not sure what you are relying on for your refutation of my hearing a difference between the two bit rates.

I do in fact hear a difference. You say you don't hear a difference.

Unless you're an engineer and can explain the acoustical difference in terms of audible hearing range and can provide me with that data I'm not sure I will be convinced out of the obvious difference I do seem hear between the two samples.

I notice the difference immediately. In fact, there is a new CD format called HDCD in which recordings are made in 18-bit. Even with a 2 bit increase I can hear a difference in sound quality on my desktop computer speakers (Midiland G2) and with the Yamaha MSP5's which go up to 40khz it is even more distinct.

What evidence do you have to refute me on this?
Sorry if I sound opinionated myself about this. I could be wrong, but I don't take lightly just the throwing around haphazardly of opinions, the world is full of opinions. I am more interested in the facts. Why should I trust your opinion over my own experience?

Besides, my question has to do with sample conversion, not "should I throw my 24/96 soundcard in the trash and sample at 44.1/16 with a Sb-live?"

Thanks for your view though. :)
 
Re: skeptical?

BM,

> I'm not sure what you are relying on for your refutation of my hearing a difference between the two bit rates. <

I'm not refuting anything. All I asked for are details of your listening test.

Many times I've had people tell me they heard an improvement of one thing or another, and then I find the test itself was defective. One time was when a friend said he heard a test of 44.1 versus 96 and he heard an improvement. But it turns out the test was someone singing into a recorder set for 44.1, and then singing again later with the switch set to 96. Ignoring that a voice doesn't have much if any content past 10 KHz. anyway, as soon as you move the mike 1/4-inch to the side, or sing with slightly more breath, the sound quality will change a lot independantly of the sample rate. In other words, the test was totally bogus.

> I notice the difference immediately. <

Again, what did you test and how?

--Ethan
 
The vast majority of humans can only hear frequencies between 20 Hz and 20 kHz. If the sampling rate is 44.1 kHz, frequencies up to 22kHz will be accurately reproduced. This is outside the human hearing zone. A big reason for the higher sample rate such as 96 kHz is the companies can use cheaper parts in their filters and such. A regular filter rolls off at 6 dB/octave. This can cause abnormalities at the high and low frequencies. If a higher sample rate is used then the manufactures can get away with using these basic filters. I'm short for time right now, but if you want a better explaination let me know.

PS. I am an engineering student and have run detailed experiments on different sampling rates using a dedicated frequency generator to produce differrent tones. The results were analyzed using a spectral analyzer. My conclusions showed no audible improvement using 96 kHz over 44.1 kHz.
 
good stuff

Thanks. That's more helpful to me. Perhaps I'm fooling myself then. Maybe I'll do a listening test again sometime soon and see if I can really tell the difference. I know I have sensitive ears. Maybe I'll have my wife play back different audio tracks without telling me what sample and bit rate they are.

There is something comforting though about recording at 96/24, I'll let you know the results of my listening test.

Thanks.
 
The true advantage to higher bit rates is headroom, this typically is realized when listening to quiet passages and judging the relative dynamics between very soft & very loud.

16-bit gives the composer 65,536 "steps" of resolution within the waveform. 24-bit increases this number to 16 million+.

When a 24-bit waveform is "dithered" down to 16-bits, the waveform is processed to smooth the difference between the "steps" to reduce the impression of artifacting at low dB levels. This is a slight improvement that, as stated, is very difficult for most people to hear vs. native 16-bit recordings.

I was solidly in the camp where I believed there was zero audible differences. Then I started working with Reason heavily which runs internally in 32-bit. Later when I started mixing down my work I was hearing a slight loss of signal in the more finely-realized parts of my music. I now feel I can hear the difference but only in specific cases and they are fairly few and far between.
 
heinz is right. You probably will hear a difference when recording at a higher bit depth. The more bits available to store information about the sampled sound, the more accurate it will be reproduced. The higher bit depth is very important if you do further digital processing. When working with a DSP sometimes you can run into bit "overflow". This is similiar to doing multiplication and carrying a number over to the next column. If there is a carry over bit, but no column for it to go then that information is lost, thereby lossing some sound quality. That is why it is important to record at the highest possible bit rate. The last step in digital mastering should be dithering down to 16 bit, if you need to put it on CD otherwise just leave it at the higher bit depth. I hope that all makes sense, it can be kinda hard to explain. Good Luck with your testing.
 
Yes

I feel like I am slowly being vindicated here on my view that higher bit rates and khz give a better quality.
 
Enough about sample rate and bit depth.

1)Try mixing down the soundfonts and samples
2) Write them as wav. files or something
3) Re-record them with sonar set-up for the sample rate and bit depth you want

Not sure if it will work, but worth a shot I think
 
I understand the difference between 24 and 16, but what about the difference between 44.1 and 96? I have never seen anyone address why quality is better when your sampling is that high.
 
sonny,

That is a tough one to explain without sitting down and drawing some pictures, but I'll give it a try.

First:
Every sound made produces some power at most every frequency. However, there is usually a peak frequency heard. For example an regular note "A" is 440 Hz. However, there are components of this "A" across the whole frequency spectrum, the other frequencies just don't have alot of power in them (very small amplitude, basically).

NExt:

When you use a sample rate of 44.1 kHz, for example, you can accurate sample frequencies up to 22.05 kHz. This is called the "Nyquist rate". Any frequency above this will subject to overlapping spectral components, this is called aliasing.

Aliasing:

Imagine a sine wave. It starts at 0 goes up to 1 down to 0, then down to minus 1 and back to 0. This is one wavelength. If this process is faster than half of the sampling rate. There will be overlap between one wavelenth to the next, hence aliasing. The digital representation will not be accurate for this wave in its entirety.

Now, it is true that a higher sampling rate will better reproduce the analog signal. But, as you have probaly heard before, human's can only hear from 20 Hz to 20 kHz (and that is for the best ears around). So, why waste vaulable storage space and processor time on accurately sampling frequecies that you won't be able to hear anyways.

Th majority of speakers can not reproduce frequencies above around 22 kHz anyways. That would mean that the drivesr for the speakers are oscillating 22,000 times a second!! Anything higher requires some very high end components and alot of dough.

Depending on the medium that the finished product will be used on (CD, Computer, DVD) there are reasons for recording at higher frequencies, but it is usually due to be able to use smaller order filters in the speakers and amplifiers. This means cheaper cost for these products.

I hope that makes sense. I could go into more detail, but it really depends on your math and electrical and signal background. Plainly speaking, don't buy into all these companies that want to charge you an extra hundred bucks or so because their gear samples at some astronomical rate. You're wasting your money.


If you want to know more pick up a "Schaum's " Digital signal processing book. You can find them at most book stores and they go into great detail about this stuff. Lots and Lots of math though. I'll try explaining some more if it wasn't clear, but I hope it helps.
 
You must log in or register to reply here.
Back
Top