The Great 96 kHz Debate....all invited!

  • Thread starter Thread starter Bob's Mods
  • Start date Start date
sampling rate vs. frequency range

What a great site. Wish I had discovered it earlier.

I'm a singer/songwriter and record very simple voice+guitar, country music demos. I record either in mono or blend a vocal and guitar mic to get a fuller guitar sound.

Here's something I've noticed: My recordings sound better on my minidisk recorder (a newer one that records in PCM and uses no data compression) than recording at 96k on my masterlink using high quality microphones (AKG 414, Neuman). (For that matter, things sound better at 44.1k on the masterlink than at 96k). Of course, the mike I use on the minidisk is not a powerhouse. 40hz-15khz, at best. But, to my ear, the minidisc recording sounds smoother and more analog than anything recorded on the masterlink.

Could it be that things sound better with a less responsive mike because there are less frequencies to sample? Are there theories about this? Literature you could point me to? Or might it be that my ears just don't like the extreme lows and extreme highs? Any opinions about this?

I want that old, intimate country singer sound. A la Hank Williams Sr.'s demos, if you're familiar with them. Of course, they were recorded on tape. But the geek in me likes digital.

What do you think? Thanks for the great forum.
 
PHILANDDON said:
Could it be that things sound better with a less responsive mike because there are less frequencies to sample? Are there theories about this?
Click to expand...

No. Converters don't sample frequencies, they sample voltage at a fixed point in time--OK, 44,100 fixed points in time per second. But the converter doesn't know how wide the frequency range is, it doesn't care. Whatever slips below the anti-aliasing filter gets captured.

I understand that jitter is more audible at high frequencies, but I really don't think that is what you are experiencing.

Or might it be that my ears just don't like the extreme lows and extreme highs? Any opinions about this?
Click to expand...

Sounds like it. Ever try a ribbon mic? That might be the ticket for you.
 
If someone can't get an amazing sound out of 24/48 with good converters, then the problem is NOT digital resolution.....[/QUOTE]


Its interesting Blue that you have selected 48 over 44dot1. I get the impression you believe sample rate is irrelevant but yet you record at 48 kHz? Why not save some work by recording directly at 44.1? Or is it you feel there is no difference between 96 & 48 but there is a difference between 44 and 48?
 
I use 44.1 OR 48, depending entirely on the project requirements -- sometimes clients bring tracks in at 44.1, so in that case I use 44.1... other times it's 48 simply because that's the "default" SR when using a BRC with an HD24/ADAT rig.

Before I got the Sony console, I would track at either 44 or 48, but mixdown at 88.2... why? Because with the Masterlink the hi-res mixdown was a no-brainer and no strain on "system resources" as such and it does make sense to record to the highest possible resolution, barring no other project requirements that dictate otherwise.

Now with the Sony console, I stay digital right through mixdown, so the chosen SR for tracking automatically becomes the mixdown rate as well... 44.1 or 48, again depending on the project requirements.
 
Does anyone know what sample rate the magor studios use? The kind of stuff thats contracted by the big boys and pushed out onto the commercial mainstream? I get the impression from Blue Bear its most likely 44 or 48.

There does seem to exist some mixed opinions regarding the use 96 kHz. Its hardly default as 44dot1 is presently.

Bob
 
Remember a couple years ago or whatever when Korn came out with a CD that was supposed to be unique and modern because they did it at 192K? It was supposed to sound like so much better, but I couldn't really tell. I don't know how many major studios are doing this now, though.
 
Bob's Mods said:
Does anyone know what sample rate the magor studios use? The kind of stuff thats contracted by the big boys and pushed out onto the commercial mainstream? I get the impression from Blue Bear its most likely 44 or 48.

There does seem to exist some mixed opinions regarding the use 96 kHz. Its hardly default as 44dot1 is presently.

Bob
Click to expand...


I heard they tracked that rapper Jay-Z's last album instrumentals at 24 Bit 48K, tracked the vocals in as 24 Bit 192 K, and mixdown to 24 Bit 88 K
 
Speaking of Neil Young......he released his "Greatest Hits" (late last year) in both CD and DVD-A formats. You can even buy both in the same case. Lemme tell ya, there's a HUGE difference in depth, resolution and overall sound quality between the two. These are recent remixes of the same analog-tape sourced material, so there's no "80s vs today" argument.

Neil has always been an outspoken critic of the way CD sounds, and has been a proponent of just about every "non-CD" format released since CD became dominant (think SACD, etc). He seems to be openly embracing DVD-A, which is something i wish more folks would do.

i urge anyone who wants to hear the difference between CD and DVD-A to go pick them up. Down By the River in 24/96.....YUM.


cheers,
wade
 
PHILANDDON said:
I want that old, intimate country singer sound. A la Hank Williams Sr.'s demos, if you're familiar with them. Of course, they were recorded on tape. But the geek in me likes digital.
Click to expand...

I'd look at different mics, reverbs and instruments. Maybe mix down to tape?
 
mshilarious said:
Umm, I'm an anti-96Ker :confused:
Click to expand...

yeah, whatever. but he is on the run, and an anti-96k-er.

here's my thoughts on the debate: why wouldn't you record at 96K if you can? hard drive space is cheap, man. and so what if it may sound slightly worse on CD, it is gonna sound way better on DVD-A or SACD.
 
I believe when the original CD standard started, they were just remastering the old tape recordings to fit into the new 16 bit/44.1 kHz standard (in the mid 80s). I do not believe digital recording was being done at that time. That didn't come along until the early 90s (but I might be wrong on this). Its a testament to the quality of tape recording to just use the tape masters to create an amazing sounding DVD as in the case of Neil Young - and how constricted the old vinyl record was, and to a lesser degree, the CD standard is. This is yet another example of the improved detail possible with wider bandwidth. I really am NOT losing my mind.

Bob
 
mcolling said:
yeah, whatever. but he is on the run, and an anti-96k-er.

here's my thoughts on the debate: why wouldn't you record at 96K if you can? hard drive space is cheap, man. and so what if it may sound slightly worse on CD, it is gonna sound way better on DVD-A or SACD.
Click to expand...

OK, but I still want to know why it sounds better. Is the Nyquist theory false? And does 96kHz automatically sound neutral or better? Could it sound worse?

Read this:

http://recforums.prosoundweb.com/index.php/t/2997/0

http://recforums.prosoundweb.com/index.php/t/3538/0

http://recforums.prosoundweb.com/index.php/t/2674/0

Or several other threads there . . .
 
The Mathmatical Truth

I'm by no means a "pro" when it comes to audio production, I've been somewhat of a weekend warrior for most of my life, and have lived through an old Studio 8 tape unit, and ADAT setup, then Digidesign Session 8, and for the last few years I've been using Cubase pretty exclusively. As I myself have been trying to come to a good conclusion on the use of "higher" sample rates the following points are important to consider:

1. A higher sample rate will capture a more accurate "picture" of the sound.
That's a mathematical fact. All "sound" is analog in nature. Meaning it is constantly fluctuating. The role of a A/D converter is to capture a snapshot of that sound pressure (translated to voltage in an audio signal) and give it a number. The sample rate determines how many times per second this happens. The more samples you can get in a given period of time, the more accurate the result. Many people refer to the law that states you must utilize a sample rate twice the highest frequency you wish you record. So... 44.1, you've got 20kHz covered right? Well... sort of. At 20kHz the most a 44.1kHz converter is going to capture is 2 samples per wave at 20kHz. So if the converter happens to capture a peak and a valley, you're not in terrible shape, but if it happens to capture another point on the wave, the result is that he converter thinks it's seeing a lower amplitude wave than actually exists. Or worse, it could capture the points on the wave which are zero and it wouldn't capture the frequency at all. So yes, 44.1 is the BARE minumum required to capture those higher frequencies, but accuracy could be greatly compromised.

2. A greater bit depth with capture a more accurate picture of the sound. Again a mathmatical fact. Before I referred to the converter assigning a number to the sample. The bit depth determines how many numbers it has to choose from. A great analogy would be colors on your computer screen. Black and white would be a 1-bit depth. Step up to 8 colors at 3-bit, at 32-bit (most modern computers use this) you get over 42 billion colors to choose from. Same thing applies to digital audio. The more numbers the converter has to choose from, the more accuratly it can capture the signal.

Now, on these to points alone it's almost a no brainer. If you have the capability, (Hard drive space isn't that much of an issue these days) why would you not record at the highest sample rate / bit depth possible?

For a long time, I've had the capability to to record at 96kHz, but I haven't. 95% of what I record ends up on CD. Which means it needs to be mastered at 44.1kHz / 16-bit. So if you're project is at 96kHz / 24-bit, how do you get there? A whole lot of math. 96 isn't even divisible by 44.1, so a "downsampler" has to make "guesses" as to what each sample should be. Same thing happens in the bitrate conversion. My experience in the past is that this downsampling process has caused substantial degredation in the quality of the audio. So much so that it far outweighed any sonic benefits of recording at a higher sample rate.

Now that was 5 years ago, and I've been pretty stuck in the mud on that issue since. More and more, I've been reading articles on the improvements in "downsampling" software. And there seem to be a few out there that do a pretty spectacular job. That's always been the linch pin for me, but If I find that downsampling has made a leap forward, I'll jump to 96kHz in a heartbeat. Not only to you yield the benefit of having the best possible quality tracks to work with, but you promise some forward compatibility with new audio formats that are sure to hit the market in the future (DVD-A for example).
 
sternobread said:
Now, on these to points alone it's almost a no brainer. If you have the capability, (Hard drive space isn't that much of an issue these days) why would you not record at the highest sample rate / bit depth possible?
Click to expand...

Did you read those links? Lavry (who doesn't have a problem with 96, but dislikes 192) argues that everything is a tradeoff. If you capture more samples, then those samples must be LESS precise. And the difference in data handling is not trivial.

As for your argument that the higher sample rate gives a more accurate reproduction, ALL of the additional detail is in ultrasonic frequencies, which are all removed by the anti-aliasing filter.
 
I generally track drums at 24/96. I tried it both ways. I use an old akai dps16 to do it. The difference that it makes is a little more sizzle in the high end and a better soundstage recording from the overheads. The other difference that it makes is halving the number of available tracks on that box. I back up the work to cd and import it into the computer to mix/dither it there in either protools or logic audio.

It's really a pain in the ass to do it that way but I like how it sounds so I put up with with it.

If you have an akai and you are doing this then I have found that the best dithering option is number 2.

I went to school and learned the math and what not.

It sounds better for that application and I will often track piano that way as well. It's less difference than going from 16 to 24 bits. But still there and very noticeable.

Another thing to think about is that you can use alot lot less steep of an anti-aliasing filter at 96k.

If your crummy soundcard supports an external clock maybe you could get a good external clock and try that out if there are woes at 96k.
 
Stern,

You've kinda stated in a different way what I mentioned early on, more bits, more precision. More bits, less error.

As far as downsampling is concerned, I use Cool Edit Pro and I am unhappy NOT with it. Although going to 44dot1 is less sweet, in my case, it still maintains much of the 96 kHz detail. But I do like the native 24/96 best of all.

Bob
 
View on 24/96 from Guy at Roland

I called Roland's tech support yesterday. Told them I have the BR1600 (a 16 bit machine), and was thinking up upgrating to the VS2400 b/c I wanted a 24/96 machine. The guy at Roland told me that while a greater bit rate makes a difference, in his view a bigger sampling rate did more harm than good if the final project was gonna end up on cd. He made an analogy to digital cameras saying that if you go from 96 to 44.1 you're merely taking a digital picture of a digital picture.

Is he right?. I didn't think sample rate conversion merely involved re-recording a 96/24 file at 44.1/16. If that is what goes on, then the only reason to record at 96, as far as I can see, would be if DVD-A catches on. Sorry to rev this up again. But what do you think of what he said? Simply put, most people agree that 24bit recording is better, but is it worth it to record at higher than 44.1 if the final product is gonna end up on cd.
 
This is religion, which means people claim that it sounds better, but all available theory sais it doesn't. ;)

Doing the mixing at 96 can in theory give you a better sound because the effects (primarily reverbs) have more datail to work with, but this is hard to prove.
 
regebro said:
This is religion, which means people claim that it sounds better, but all available theory sais it doesn't. ;)

Doing the mixing at 96 can in theory give you a better sound because the effects (primarily reverbs) have more datail to work with, but this is hard to prove.
Click to expand...

just listen, the reverbs tails are much longer i guess doubled
 
This guy's articles, Part 1 and Part 2 are very good treatises on digital signal theory.

To answer the question, you should record at the same sampling frequency that your final product will be. If your gonna make audio CDs stick with 44.1 kHz.
 
You must log in or register to reply here.
Back
Top