Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: M-Audio 24/96 (Read 8834 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

M-Audio 24/96

Reply #25 –
Quote
You seemed to suggest that recording a 24-bit audio signal into a 32-bit audio file and converting it to 16bits was a bad thing to do, and that recording straight to 16-bits would be better.

No no no!

No, not that it's a bad thing to do -- just that it's counterintuitive that one could get better quality from:

[A] Recording to 32-bit float, then downconverting to 16-bit with dither and noise shaping, then saving the file, vs.

Recording to 16-bit, then saving the file.

[A] involves a (fairly major) extra processing step, involves no steps at all after recording, and trusts the card to do 16-bit recording properly as a good quality card should do.  Therefore, I couldn't understand why everyone was suggesting [A] as the method to use for quality.

Yes, I suggested that I might increase the volume in software, but I wouldn't if I were recording in 16-bits.  To me this isn't mental masturbation at all, but a matter of practicality (fitting quality issues in with other concerns).  The issue of vinyl aside, it could apply to *any* recording done through the 24/96's analog-in jacks.

M-Audio 24/96

Reply #26 –
Quote
[A] Recording to 32-bit float, then downconverting to 16-bit with dither and noise shaping, then saving the file, vs.

Recording to 16-bit, then saving the file.

[A] involves a (fairly major) extra processing step, involves no steps at all after recording.

Therefore, as dithering permits keeping more 'resolution' than a 16bit would offer in the first place, A is better

M-Audio 24/96

Reply #27 –
Quote
Quote
[A] Recording to 32-bit float, then downconverting to 16-bit with dither and noise shaping, then saving the file, vs.

Recording to 16-bit, then saving the file.

[A] involves a (fairly major) extra processing step, involves no steps at all after recording.

Therefore, as dithering permits keeping more 'resolution' than a 16bit would offer in the first place, A is better 

16 bits is 16 bits, how am I keeping more resolution than simply recording in 16 bits in the first place?  Can one cram more than 16 bits of dynamic range into 16 bits? 

I still fail to see the "benefit" (unless I were to do a lot of processing on the file), and in my opinion nobody's satisfactorily explained why recording at a higher bit depth and then downsampling with dither is better.  Bryant's suggestion re: truncation remains unverified and is purely a guess lacking any hard data about the 24/96 and 16-bit recording.  It's a bit surprising to hear "explanations" completely lacking in proof around here, I thought everyone knew already how they just confuse and obfuscate.

Edit -- I've contacted M-Audio's tech support on this matter, perhaps they'll be able to clear it up.

M-Audio 24/96

Reply #28 –
Quote
16 bits is 16 bits, how am I keeping more resolution than simply recording in 16 bits in the first place?  Can one cram more than 16 bits of dynamic range into 16 bits? 

I still fail to see the "benefit" (unless I were to do a lot of processing on the file), and in my opinion nobody's satisfactorily explained why recording at a higher bit depth and then downsampling with dither is better.  Bryant's suggestion re: truncation remains unverified and is purely a guess lacking any hard data about the 24/96 and 16-bit recording.  It's a bit surprising to hear "explanations" completely lacking in proof around here, I thought everyone knew already how they just confuse and obfuscate.

Edit -- I've contacted M-Audio's tech support on this matter, perhaps they'll be able to clear it up.

To repeat a link given by 2Bdecided, which is excellent:

http://www.mtsu.edu/~dsmitche/rim420/readi...420_Dither.html

If you record straight to 16 bits, you just get straight 16 bit resolution.  If you record to 24 bits and dither down to 16, you can, in theory, hear stuff below 16-bit resolution.  It is iffy that you will be able to hear the improvement, though (mental masturbation, as it is put).  On the other hand, one of the major points about owning a 24-bit soundcard is the conceit that one is getting better than 16-bit resolution.

ff123

M-Audio 24/96

Reply #29 –
Quote
If you record straight to 16 bits, you just get straight 16 bit resolution.  If you record to 24 bits and dither down to 16, you can, in theory, hear stuff below 16-bit resolution.  It is iffy that you will be able to hear the improvement, though (mental masturbation, as it is put).  On the other hand, one of the major points about owning a 24-bit soundcard is the conceit that one is getting better than 16-bit resolution.

Well, not to me.  Mostly I bought the M-Audio 24/96 for the much better specs regarding both resistance to RFI/EMI (noise floor) and lack of weirdo harmonics that affect cheaper cards.  For example, a certain high pitched whine (probably due to EMI from system fans) that affected my SB PCI128 at about -78dB is completely absent with the 24/96.  Also, the PCI128 added a sort of "brightness" to recordings from analog in which definitely isn't there either with the 24/96.

I do understand the benefits of editing the file in 24 or 32-bits and downconverting, although practically speaking, a plugin I really want to use (search and destroy on individual clicks/pops) only works in Sound Forge v4.5 (limited to 16 bits) and I'll probably keep using it.  For initial recording and maybe volume changes, I suppose I'll be going to a different program and recording in 32-bit float.  Haven't made a final decision yet.

Thanks much for the input...

M-Audio 24/96

Reply #30 –
My stupid guess...  (okay i read more than the last post and this is already said i guess )

If you record @16bit and then playback the same file its fine.  I'm doubting anyone could ABX 16bit vs 24bit dither to 16.

But lets say you record at 16bits and your signal is low you don't ever put all the bits to use, you don't have 16bits of resolution, maybe 14 or something.  Had you recorded at 24 you could amplify the sound and then dither down to 16, you'd end up with 16bits of resolution.  What happens if you try to record @ 16bit and make use of all the resolution..  can happen.  WHat happens then if you clip.  Had you recorded at 24 it wouldn't of clipped and you could scale it correctly and then dither do 16bit.

Reason for recording at 24bit would insure that you get 16bits of real resolution in the end.  If you record at 16 at best you can get 16, if anything is wrong you don't get it.

M-Audio 24/96

Reply #31 –
Quote
If you record @16bit and then playback the same file its fine.  I'm doubting anyone could ABX 16bit vs 24bit dither to 16.

http://ff123.net/24bit/24bitanalysis.html

There are some real golden ears out there, so it is possible.

ff123

M-Audio 24/96

Reply #32 –
Quote
Quote
If you record @16bit and then playback the same file its fine.  I'm doubting anyone could ABX 16bit vs 24bit dither to 16.

http://ff123.net/24bit/24bitanalysis.html

There are some real golden ears out there, so it is possible.

ff123

Interesting:

"2. In contrast to the group results, two listeners demonstrated that they could distinguish between the two 16-bit files, both indicating a preference for the truncated 16-bit file."

In other words, these two listeners preferred truncation 24 --> 16 rather than dithering 24 --> 16?

P.S. I'm still liking the idea better of just recording in 16 bits using Sound Forge, and minimal or no processing.  I have this unshakeable idea that adding more processing steps (of which dithering 24 -> 16 is a major one) just f&x*rs sound quality while adding only 'theoretical' benefits... especially in the case of recording vinyl, where noise floor is at -80dB before even putting the needle on the record.

M-Audio 24/96

Reply #33 –
Quote
Interesting:

"2. In contrast to the group results, two listeners demonstrated that they could distinguish between the two 16-bit files, both indicating a preference for the truncated 16-bit file."

In other words, these two listeners preferred truncation 24 --> 16 rather than dithering 24 --> 16?

Yep.  That is interesting, isn't it?

But in general, that analysis said that 13 bits were enough for this sample and group of listeners!  I seem to recall that 14 bits was at one time being considered for some digital standard, I don't know whether it was CD or not.

ff123

M-Audio 24/96

Reply #34 –
Quote
But in general, that analysis said that 13 bits were enough for this sample and group of listeners!  I seem to recall that 14 bits was at one time being considered for some digital standard, I don't know whether it was CD or not.

ff123


OFFTOPIC: 14bit was advocated by Philips to be enough for CDDA. However, Sony had asked for an increase in the planned disc size (initially 60 mins). More important, Sony had persuaded Philips that the planned 14-bit word length was insufficient to achieve the desired sound quality. SONY also came up with the error-correction code and was a forceful advocate for 16-bit word lengths.

First DACs produced by Philips featured a 14bit D/A conversion anyway (most CD players before 1985-86 used Philips DACs).

As for the length, ~74 mins (coincidence?) is enough to fit Beethoven's 9th Symphony conducted by Herbert von Karajan.

Tip: Search for 14 bit enough for cd standard 74 minutes on Google...

edit: typos

M-Audio 24/96

Reply #35 –
Re: the system tray:  In 2K and XP there's a nifty little system tray applet.  Double-clicking it just runs the normal Delta control panel applet, but it's awfully handy.  Don't know why it's not in the 9x drivers...

M-Audio 24/96

Reply #36 –
Quote
Re: the system tray:  In 2K and XP there's a nifty little system tray applet.  Double-clicking it just runs the normal Delta control panel applet, but it's awfully handy.  Don't know why it's not in the 9x drivers...

They must have ceased development of the VxD drivers.  No surprise there, and I'm not complaining.  It is I who have decided to remain with older OS software, while the rest of the world moves on to bigger, flashier and ostensibly more powerful and stable. 

Let's just I don't care for the direction MS is headed with their latest OS software (or shall we say, all of their software).  Linux is definitely in my future, and that future is by no means distant.

M-Audio 24/96

Reply #37 –
Quote
As for the length, ~74 mins (coincidence?) is enough to fit Beethoven's 9th Symphony conducted by Herbert von Karajan.


I've heard this before, but I still don't know.
I have a copy of von Karajan and the Berlin Philharmoniker performing Symphony no. 9 in D Minor, ("Choral" og. 1977 Polydor - on Deutsche Grammophon 415 832-2) and it only runs 66'54" 

M-Audio 24/96

Reply #38 –
Quote
P.S. I'm still liking the idea better of just recording in 16 bits using Sound Forge, and minimal or no processing.  I have this unshakeable idea that adding more processing steps (of which dithering 24 -> 16 is a major one) just f&x*rs sound quality while adding only 'theoretical' benefits... especially in the case of recording vinyl, where noise floor is at -80dB before even putting the needle on the record.

Despite that, you will get better quality doing the other suggested way. 32 bit floating point processing error residuals are far beyond 16-bit dynamics, and far, far beyond out hearing capabilities. That's how they work at studios. Conversion to 16 bit is considered a degradation of quality, and for that reason, must be performed as a final step, and done with best methods (using dither).

M-Audio 24/96

Reply #39 –
Quote
Quote
P.S. I'm still liking the idea better of just recording in 16 bits using Sound Forge, and minimal or no processing.  I have this unshakeable idea that adding more processing steps (of which dithering 24 -> 16 is a major one) just f&x*rs sound quality while adding only 'theoretical' benefits... especially in the case of recording vinyl, where noise floor is at -80dB before even putting the needle on the record.

Despite that, you will get better quality doing the other suggested way. 32 bit floating point processing error residuals are far beyond 16-bit dynamics, and far, far beyond out hearing capabilities. That's how they work at studios. Conversion to 16 bit is considered a degradation of quality, and for that reason, must be performed as a final step, and done with best methods (using dither).

I have to say that the kind of crap (to my ears) coming out of many studios these days doesn't inspire any more confidence in the method... 

M-Audio 24/96

Reply #40 –
Quote
I have to say that the kind of crap (to my ears) coming out of many studios these days doesn't inspire any more confidence in the method... 

Still, it's the best quality way, from any point of view. Actual recordings sound bad from other reasons. Ask anyone that knows about digital audio, he'll say the same thing.

M-Audio 24/96

Reply #41 –
Listening to dithering (for pure curiosity reasons): open foobar, switch output data format to 8bit. Then listen to a favorite song w/ dither checked/unchecked.

I guess it's safe to assume that dithering increases resolution at the expense of slightly higher noisefloor (not noticeable when using ATH noise shaping in 16 bit/higher IMO).

M-Audio 24/96

Reply #42 –
Quote
Listening to dithering (for pure curiosity reasons): open foobar, switch output data format to 8bit. Then listen to a favorite song w/ dither checked/unchecked.

I guess it's safe to assume that dithering increases resolution at the expense of slightly higher noisefloor (not noticeable when using ATH noise shaping in 16 bit/higher IMO).

My phono preamp + cartridge puts out basically white noise (along with a bit of 60 cycle hum) at -80dB before the needle is ever put onto the record.

I wonder if in this case, the white noise "dither" from the preamplifier is more than good enough for recording at 16-bits?  The only difference is probably in processing of the audio.

I'm gonna assume it's "safe" to record either way (32float-->16 in Cool Edit or just 16 bits in Sound Forge), and use 32-bit float when I need to do some processing of the audio.  Trouble is, I have to remove clicks/pops first before most other processing (except possible initial volume boost), and the manual plugin I prefer for that only works in Sound Forge 4.5 and below (16-bit only).  Ahh well, practicality concerns + plugin preferences trump theoretical sound quality... 

If I could find another effective "manual search and destroy" click/pop plugin, the issue would be moot.  Unfortunately, all the others seem to be automatic and not very effective.