OT: Who here has recorded at 32/192?

[Frippertronix]

I posted this in hardware, but I hang out here more so I thought I'd ask this crowd as well.

M-Audio has an affordable 64/192 card now, so I'm wondering if it's worth upgrading.

My DAW (Tracktion 2) can support it. Problem is, many of my effects can't.

Anyone tried recording at higher resolutions than 24/96, and if so, did you feel it made a big improvement?

Thanks much,
Dave

[Kim Lajoie]

I'd just like to point out that there's no difference between recording at 24 bit and at 32 bit.

1) ADCs are no longer limited by bit depth. For modern 24 bit ADCs, the weakest link is the analogue components.

2) Typically, 32 bit (in audio) refers to 32 bit floating point. It has 24 bits "worth" of resolution, and the remaining 8 bits are a bit like a lossless multiplier, that allows those 24 bits to be fully used at any dynamic level. Either way, this doesn't help in the recording process, because the ADCs capture at 24 bit integer.

Of course, I would very much like to be re-educated.

Forever,




Kim.

[Spe3D]

I get better results ‘recording’ at 24-bit 44.1 than at 192 – I have tried both but I think it’s the quality of sound card and the type of material recorded that makes the difference, maybe its better to record at the highest bit depth but keep the sample rate as whatever the end product is going to be.



Best regards,

Spe3d

:O)

[kritikon]

I'm with him.....I record at 24/44.1 externally. Anything internally rendered is at 32/44.1

I very briefly played with high samplerates but gave it up almost straight away - the extra CPU it gobbles up just makes my DAW impractical to get anything done at those sample rates. I have no headroom for scratching out a song with. There may be an audio benefit with high samplerates, but I'm not even going to argue about that one - for me it just doesn't work. I'm more than happy with 44.1 quality. I don't have a particularly slow PC, although it's certainly not bang up-to-date either...it's an AMD 2.5GHz CPU and on some VSTi I'd struggle to do more than 4-note chords with a single FX at 96KHz, never mind 192! If I even comtemplated 192, I'd have to do everything in mono probably, render absolutely everything as I go and have no leeway to build up midi channels before commiting myself to audio.

The thing is - do you think your audio quality is bad at the moment? I don't mean your mixing, choice of FX etc - your actual recording quality. If you're happy with it, why spend on another card just because it's available. One day I'm sure it will become the standard, but commercial standard is still set at 44.1/16 for CD and is likely to stay that way for several years yet - I can see DVD possibly taking over in the future, but not yet. And until that day, why flog your PC to death until they become powerful enough to handle that level? PCs are powerful enough now to handle an awful lot at 24 or 32bit/44.1 - that's a great advantage, so why throw it away?

[Left Headphone]

I get better results ‘recording’ at 24-bit 44.1 than at 192 – I have tried both but I think it’s the quality of sound card and the type of material recorded that makes the difference, maybe its better to record at the highest bit depth but keep the sample rate as whatever the end product is going to be.



Best regards,

Spe3d

:O)

Me 2

C U Later,

Lefty

[Frippertronix]

My bad---I thought top of the line s/c's actually recorded at 32 now.

I do perceive a bit of a change doing 96 instaed of 48. The overall sound, to me, is a bit less "edgy" in the "digital" sense.

Dave

[Frippertronix]

1) ADCs are no longer limited by bit depth. For modern 24 bit ADCs, the weakest link is the analogue components.
.

Can you elaborate. What do you mean by "ADC's are no longer limited"?

[bitcrusher]

1 bit = 1 signal doubling = 6 dB...
at 24 bits in a cheap soundcard, you're already losing a couple of bits to noise on the ADC's.

32 bits would just mean you have high-resolution noise captured.

Sample Rate is more important (especially on stereo signals as you can percieve phase information beyond the audible frequency spectrum).

I did a blindfold shootout recently of drums tracked to tape, ProTools HD 44.1, 96, and 192k in a nice studio (Record Plant). The difference from 44.1 to 96 is very noticable on stereo transient-heavy material. The audible difference from 96 to 192k is splitting hairs, for sure. I knew the tape when I heard it every time, but it was only a very subtle emphasis right in the transient (crack of a snare) that could make me distinguish it.

Obviously, by the time some kid is downloading your song as a 128k mp3, none of this matters at all.

[Frippertronix]

Sample Rate is more important (especially on stereo signals as you can percieve phase information beyond the audible frequency spectrum).

I did a blindfold shootout recently of drums tracked to tape, ProTools HD 44.1, 96, and 192k in a nice studio (Record Plant). The difference from 44.1 to 96 is very noticable on stereo transient-heavy material. The audible difference from 96 to 192k is splitting hairs, for sure. I knew the tape when I heard it every time, but it was only a very subtle emphasis right in the transient (crack of a snare) that could make me distinguish it.

Obviously, by the time some kid is downloading your song as a 128k mp3, none of this matters at all.

Agreed on your last point---but hey, I'm making timeless music that will be celebrated in 250 years, much as the music of J.S. Bach is today.

Joking aside, I like to use the higher bit depth/sample rate to cut masters, since, especially with the huge popularity of iPod and friends, I think it won't be too long before the totalitarian 16/44.1 format will not be a brick wall for personal music systems. Might as well make your master recordings have some longevity in terms of remaining up to date as consumer standards go up in the future.

[Frippertronix]

I'd also like to say that, while it's obvious that very high transients are going to resolve better by not being chopped off as you increase sample rates, you also get, from my understanding anyway, a smoother curve represented by increased sampling, which suggests something closer to an analog representation of the waveform. Pardon my ignorance if that's way off, but I've read articles that seemed to be saying that.

That brings up a question: can someone explain what oversampling accomplishes?

[ericj23]

oversampling has two functions -it essentially adds extra data points into the stream - these can just be as the samples around it or they can be shaped to be a kind of join the dots between the "real" data points

one purpose is to provide sufficient data points that the aliasing resulting from the digital filter, which is needed to remove any signals at over the nynquist frequency (and probably a bit lower), are well outside the human hearing range.

the other thing is something i dont remember clearly

[donkey tugger]

Probably for widdly c**ts I'll wager.

[mandolarian]

Frippertronix, it's way off. More dots don't make for better accuracy. RTFM. The Nyquist Sampling Theorem Manual. Download and read this from Mr. Sampling Theory Expert, Dan Lavry. http://www.lavryengineering.com/documen ... Theory.pdf
WARNING: Document contains Mathematical functions. If you are allergeric to complex formula or are a guitarist, look only at the pretty pictures.

Your ADC is already oversampling in the range of 64 to 512x and the point is to make it easier for anti-aliasing filtering and flatter amplitude response. And greater sex appeal.

[Frippertronix]

WARNING: Document contains Mathematical functions. If you are allergeric to complex formula or are a guitarist, look only at the pretty pictures.

.

WARNING 2: If you are a lead vocalist, don't even bother with the pictures. They contain no examples of ample female body parts.

[Frippertronix]

Frippertronix, it's way off. More dots don't make for better accuracy. RTFM. The Nyquist Sampling Theorem Manual. Download and read this from Mr. Sampling Theory Expert, Dan Lavry. http://www.lavryengineering.com/documen ... Theory.pdf
Your ADC is already oversampling in the range of 64 to 512x and the point is to make it easier for anti-aliasing filtering and flatter amplitude response. And greater sex appeal.

Thanks for the link, I've glanced at it and will read more of it later.

The problem with stuff like this is that it is:

a) The product of the minds of engineers.

and

b) Seems to be saying that current digital technology is already equal to or better than the higher quality analog technology in terms of accuracy, "naturalness", etc., of sound reproduction and that is a big, nasty, raging debate that I doubt will be completely put to rest by articles (well thought out though they may be) such as this one.