Twenty Four / 96?

[munchkin]

I've just built a new PC with a bit more power and disk space. I'm trying to use my RME Multiface soundcard at 24bit/96kHz but it's crackling a wee bit on occassions. Also some of my plugins can't cope with 24/96 (Spin Audio 3D Delay/Chorus so far.)

Does anyone else record in 24/96? I'm interested in hearing about ways to optimise WinXP for 24/96. Sound quality is audibly better at 24/96 but is 24/48 a worthwhile setting for recording if 24/96 causes crackling? (Some of my Powercore plugins only work at 48kHz.) Or should I stick with 16/44? Most of my samples are 16/44.

[rp314]

Sound quality is audibly better at 24/96

If you find this to be true, then I would think it makes sense to try to work out the crackling issues. I use 24/48 or 24/44.1 depending on the project but my hearing has been affected by life style/environmental factors.

[munchkin]

Sound quality is audibly better at 24/96

If you find this to be true, then I would think it makes sense to try to work out the crackling issues. I use 24/48 or 24/44.1 depending on the project but my hearing has been affected by life style/environmental factors.

Do you think 24/48 gives much better sound? In the end it's going to be turned into 16/44 or mp3 anyway. I'm not sure...

[Jason Brian Merrill]

if you do a search on the subject i feel the general consensus is that 24 bit is worth it, 96k is debateable. However, running your vstis at 96k could prevent aliasing....

[james0tucson]

Do you think 24/48 gives much better sound? In the end it's going to be turned into 16/44 or mp3 anyway.

Says who?

I think the dynamic range argument is different if you're recording flute and piano in a conservatory room, than if you're recording guitar from a high-gain guitar amp. If you're recording metal or loud dance music, why would you be worried about maximizing your dynamic range anyway? The idea of dynamic range varies dramatically, because some styles of music try to be perceived as "loud as possible", whereas other styles are shooting for "quiet as possible." If you are going for the latter, where you want the maximum amount of room in the dynamics domain between the solo ppp flute and the full fff orchestra, you can make the case for recording in 24bits.

As for the frequency response argument, I think it is unfortunate that it gets lumped in with the dynamic range argument. It's *much* easier to show examples where headroom in the dynamics domain makes sense, whereas there hasn't been a single bit of evidence that anyone can actually hear the difference between a 96kHz recording and a 48kHz recording, at least not in any kind of a controlled test.

If you can actually produce such evidence, it would be quite important. Unprecedented, actually.

[rp314]

As for the frequency response argument, I think it is unfortunate that it gets lumped in with the dynamic range argument. It's *much* easier to show examples where headroom in the dynamics domain makes sense, whereas there hasn't been a single bit of evidence that anyone can actually hear the difference between a 96kHz recording and a 48kHz recording, at least not in any kind of a controlled test.

You know, I've noticed recently in various forums that the distinction is becoming more and more part of the conventional wisdom, despite the continuing debate which many of us still have some interest in.

I know of no one who has the choice yet continues to record at 16 bits but more and more folks are using 44.1 or 48K for the practical as well as possibly "objective" reasons.

[noetic]

I know of no one who has the choice yet continues to record at 16 bits but more and more folks are using 44.1 or 48K for the practical as well as possibly "objective" reasons.

Yeah, I've definately seen that trend. Seeing as most hosts and plugs operate internally in 32-bit depth, there doesn't seem to be much of a performance sacrifice, if any, of increasing the word length to 24. Storage needs are about 35% greater than 16 bit at the same sample rate (I assume), but storage is relatively cheap. However the massive differences in both storage and processor load of higher sample rates has priced many out of the market. It's true that the jury's still out on the sonic benefits of greater resolution in the time domain. But even in the amplitude domain, I don't find anybody recording at 24 bits complaining about its S/N capabilities. Seems today's complaints tend to go beyond word length and sample rates (like the quality of clocking, analog stages, digital summing, etc).

I'm surely not any more prescient than the next guy, but I can imagine that 20 years from now, we could be still recording at 24/44kHz, but our converters and digital mixers will sound significantly better.

[hibidy]

Ok, no flaming, just wanna give some info.

24/96 is the standard now.......there are many reasons but I'm not sure I want to type all of them right now.........mabye a few at a time.

MUNCHKIN...please let me know your buffer size

[hibidy]

The first and probably the most important reason why 96 sampling rate is highly sought is because it gives you automatic improved latency.

Switch any host/card from 44.1/48/88.2/96....etc with a 512 buffer setting, and you will always see the latency increase immdiatly. For example, I run 24/96 with a 256 buffer and it's just sick how low the latency is. It puts a bit extra on my cpu's but it worth it.

[munchkin]

This is interesting reading. Is there is a bit rate/frequency/disk space conversion tool available or information on the internet? I did a search but couldn't find anything.

My buffer size is 256 (6ms). I tried to go lower but the Multiface canne take it captain!

[hibidy]

256 is really low for 24/96

but unless you have inadequate ram/cpu/DRIVERS etc.........that should be ok........

[munchkin]

256 seems quite low. Is that usually the lowest most audio cards run at or do some people run below 6ms?

Above what level is latency usually audible?

[rp314]

24/96 is the standard now.......there are many reasons but I'm not sure I want to type all of them right now.........mabye a few at a time.

I can't speak for anyone else but as I stated above, if you can hear a difference, believe that it's superior, and have no storage or performance issues, then knock yourself out. You're not going to hurt anyone by going 96KHz and plenty of folks are making great music doing just that.

My testimony just reflects how I see things and what I'm doing but it's really not that big a deal.

[munchkin]

I've gone for 24/48 at 256 (6ms) for now. It seems to work well with the Multiface and probably won't use as much disk space as 24/96. I think 24/96 was pushing the Multiface to its limits.

[BertKoor]

If it crackles, first try what happens if you double your ASIO buffer size.

Is there is a bit rate/frequency/disk space conversion tool available or information on the internet? I did a search but couldn't find anything.

It's too simple:

Code: Select all

FileSize = RecordingLength * ChannelCount * BitRate * ByteSize

RecordingLength is measured in seconds, and the outcome is in bytes. Divide by (1024*1024) for megabytes.
ChannelCount is usually 2 (stereo)
BitRate is 44.100 or 48.000 or 96.000 (or whatever)
ByteSize is 2 for 16bits and 4 for 32bits.

NB: most hosts actually write 24bits as 32bits! So recording in 24bits makes files twice as large.

A buffer size of 256 at 96k makes the latency the same as a buffer size of 128 at 48k, but it's far more demanding on your system.