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 8836 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

M-Audio 24/96

Got it in the mail today, plugged 'er in, drivers installed no problem...

Oops!  Win98SE is insisting on putting both sound cards (this one and my PCI128) on IRQ 11, along with my 3Com Etherlink adapter.  DOH... Disaster, my son. 

So anyway, I go into motherboard settings and manage to force the PCI128 onto IRQ5.  Ahhh... now we're cranking.

The 24/96's H-W interface seemed pretty easy to figure out, but... WTF, no handy-dandy system tray icon?  I actually have to reload this mixer thing each time I want to use it... when was this thing designed, back in 1993?  Crap.  Regular software volume controls on programs like WinAMP are, of course, non-functional.  Double-crap.

OK fair enough, I can live with that I guess...

DirectX is defaulting to the PCI128, and I suppose the only way to change this would be to change Windows Mixer settings (Control Panel/Multimedia) to the 24/96... which would invalidate the reasons I left the PCI128 in in the first place (system beeps, etc).  DOH!!

Gosh, so the 24/96 sounds pretty good... nothing monstrously and blatantly noticeably better than the PCI128, but very good (maybe I need better headphones).  Too bad the monitor mixer only allows line-in levels to be cut and not boosted... that's a pain when trying to maximize recording levels from LP's without clipping.

Questions, comments, inputs?  Can anyone help with the DirectX thing, and tell me what software records in 24 bits?  Certainly not Cool Edit Pro v1.2.

M-Audio 24/96

Reply #1 –
Well, I have the Delta Dio 24/96, which is almost the same except it has coax and optical digital I/O and no analog input (I use an external ADC).

The reason they didn't put in recording level control is that it's kind of expensive to do with high quality; you have to use a little DAC and a high-quality voltage controlled amplifier. On a cheap card they could run the DAC at some high level and then do the volume in the digital domain, but that's a very poor idea with respect to quality. If you are using a preamp then you can simply use the output that is affected by the volume control and make sure you have the tone controls flat (or defeated). Unfortunately, I vaguely remember you talking about stand-alone phono preamps (which would not have adjustable output). Even if you had to run through another preamp for volume, that would introduce far less distortion than what they could put on the card to do the same thing (unless they spent a lot of money).

You can leave the control panel thing minimized when you are needing it often; that's what I do.

That card is optimized for the highest possible audio quality at a very low price and the compromises that they made actually seem pretty reasonable to me.

Hey, you changed your post! You don't want to use their volume control to turn down the record level because that's in the digital domain (post ADC) that I mentioned above. Use external control and max the levels in the control panel. I use CoolEdit 2000 for 24/96 stuff and it works great; just bought the click and pop plugin. IIRC, there's even a nice WavPack filter for Cool Edit now... 

M-Audio 24/96

Reply #2 –
Quote
Hey, you changed your post! You don't want to use their volume control to turn down the record level because that's in the digital domain (post ADC) that I mentioned above. Use external control and max the levels in the control panel. I use CoolEdit 2000 for 24/96 stuff and it works great; just bought the click and pop plugin. IIRC, there's even a nice WavPack filter for Cool Edit now... 

So under "Recording" in an app, I want to select "PCM-IN 1/2 Delta-AP" rather than "Mon.Mixer Delta-AP" ? 

Are you sure about this?  The reason I ask is that the digital mixer is on-card and not in software... the software merely controls the card.  What will I be losing in terms of sound quality by using the mixer output to record from rather than "directly" from the inputs?

BTW, I'm certainly gonna pump the *output* through the mixer -- otherwise, there's absolutely no way to adjust volume when listening with headphones.  No other choice.

M-Audio 24/96

Reply #3 –
Quote
Quote
Hey, you changed your post! You don't want to use their volume control to turn down the record level because that's in the digital domain (post ADC) that I mentioned above. Use external control and max the levels in the control panel. I use CoolEdit 2000 for 24/96 stuff and it works great; just bought the click and pop plugin. IIRC, there's even a nice WavPack filter for Cool Edit now... 

So under "Recording" in an app, I want to select "PCM-IN 1/2 Delta-AP" rather than "Mon.Mixer Delta-AP" ? 

Are you sure about this?  The reason I ask is that the mixer is on-card and not in software... the software merely controls the card.  What will I be losing in terms of sound quality by using the mixer output to record from rather than "directly" from the inputs?

The mixer is on the card, but it's digital (which is why all the inputs and outputs have to run at the same sample rate). So, if you have the volumes all maxed, then the output of the ADC and the mixer should be identical. But the analog input goes directly into the ADC and so by lowering the volume you are just scaling down the values. If the input to the ADC is causing it to clip, then lowering the volume on the mixer won't fix it (it will just clip at some lower level).

I am basing this on my card, however. I guess they could have changed some things; isn't this addressed in the manual?

M-Audio 24/96

Reply #4 –
Quote
The mixer is on the card, but it's digital (which is why all the inputs and outputs have to run at the same sample rate). So, if you have the volumes all maxed, then the output of the ADC and the mixer should be identical. But the analog input goes directly into the ADC and so by lowering the volume you are just scaling down the values. If the input to the ADC is causing it to clip, then lowering the volume on the mixer won't fix it (it will just clip at some lower level).

I am basing this on my card, however. I guess they could have changed some things; isn't this addressed in the manual?

OK, I understand that the mixer on the card is digital, but not understanding anything else you're saying here.  So it's digital -- in what way exactly does it affect sound quality to run things through the on-card digital mixer first (perhaps scaling the values down) before recording to the hard drive?  Artifacts, what?

M-Audio 24/96

Reply #5 –
Quote
Quote
Quote
Quote
Hey, you changed your post! You don't want to use their volume control to turn down the record level because that's in the digital domain (post ADC) that I mentioned above. Use external control and max the levels in the control panel. I use CoolEdit 2000 for 24/96 stuff and it works great; just bought the click and pop plugin. IIRC, there's even a nice WavPack filter for Cool Edit now... 

So under "Recording" in an app, I want to select "PCM-IN 1/2 Delta-AP" rather than "Mon.Mixer Delta-AP" ? 

Are you sure about this?  The reason I ask is that the mixer is on-card and not in software... the software merely controls the card.  What will I be losing in terms of sound quality by using the mixer output to record from rather than "directly" from the inputs?

The mixer is on the card, but it's digital (which is why all the inputs and outputs have to run at the same sample rate). So, if you have the volumes all maxed, then the output of the ADC and the mixer should be identical. But the analog input goes directly into the ADC and so by lowering the volume you are just scaling down the values. If the input to the ADC is causing it to clip, then lowering the volume on the mixer won't fix it (it will just clip at some lower level).

I am basing this on my card, however. I guess they could have changed some things; isn't this addressed in the manual?

OK, I understand that the mixer on the card is digital, but not understanding anything else you're saying here.  So it's digital -- in what way exactly does it affect sound quality to run things through the on-card digital mixer first before recording to the hard drive?  Artifacts, what?

Well, first you should verify that what I'm saying is correct with a little experiment. Set the volumes all at maximum and provide an analog input to the card that just barely causes clipping (0 dB). Then lower the mixer volume until it's only going to a much lower level, like -20 dB. Now, increase the level of the source going to the card and see if you can get it to clip again (at 0 dB). If you can then I'm all wet and you should ignore me now and in the future. 

If the level won't go up much past -20 dB even when you crank the volume into the card then I'm right and you should also see why this is not good. First, you can't really tell if the ADC is clipping because it won't go up to 0 dB. Second, if the ADC is not clipping (which you don't want anyway) then why would you want to lower the volume digitally and destroy information? You want to have the ADC operating as close to clipping to insure it never will (-10 dB?) and then use that data directly into your software (i.e. by having the volumes maxed or bypassing the mixer for recording).

BTW, we need to stop nesting these quotes! 

M-Audio 24/96

Reply #6 –
Quote
Well, first you should verify that what I'm saying is correct with a little experiment. Set the volumes all at maximum and provide an analog input to the card that just barely causes clipping (0 dB). Then lower the mixer volume until it's only going to a much lower level, like -20 dB. Now, increase the level of the source going to the card and see if you can get it to clip again (at 0 dB). If you can then I'm all wet and you should ignore me now and in the future. 

If the level won't go up much past -20 dB even when you crank the volume into the card then I'm right and you should also see why this is not good. First, you can't really tell if the ADC is clipping because it won't go up to 0 dB. Second, if the ADC is not clipping (which you don't want anyway) then why would you want to lower the volume digitally and destroy information? You want to have the ADC operating as close to clipping to insure it never will (-10 dB?) and then use that data directly into your software (i.e. by having the volumes maxed or bypassing the mixer for recording).

BTW, we need to stop nesting these quotes! 

OK, you're right (except that the volume *does* go up past -20dB according to Sound Forge)... however, running things through the digital mixer allows me to conveniently decrease recording levels when desired, without removing my ass from the computer chair or doing software normalization afterward. 

I'm still wondering exactly what quality detriments there are from running the ADC output through the on-card digital mixer before recording... if you can't specify something then maybe you *are* all wet  .

If you do know, then how about running the DAC through the digital mixer before the analog output?  Clearly, I need to adjust volume when listening via headphones, lest I go deaf.  If you can help with these specific questions I'd appreciate it, since they are what matter to me.

M-Audio 24/96

Reply #7 –
Quote
OK, I understand that the mixer on the card is digital, but not understanding anything else you're saying here.  So it's digital -- in what way exactly does it affect sound quality to run things through the on-card digital mixer first (perhaps scaling the values down) before recording to the hard drive?  Artifacts, what?

32 bit for internal mixing, so don't worry about any quality loss.

The mixer thingy (app) is due to the fact that (at least in theory), M-Audio drivers bypass windows' Kmixer.

Would be wise to download the latest beta driver from M-Audio (29x12 Pro if I remember correctly).

M-Audio 24/96

Reply #8 –
Quote
Quote
OK, I understand that the mixer on the card is digital, but not understanding anything else you're saying here.  So it's digital -- in what way exactly does it affect sound quality to run things through the on-card digital mixer first (perhaps scaling the values down) before recording to the hard drive?  Artifacts, what?

32 bit for internal mixing, so don't worry about any quality loss.

The mixer thingy (app) is due to the fact that (at least in theory), M-Audio drivers bypass windows' Kmixer.

Would be wise to download the latest beta driver from M-Audio (29x12 Pro if I remember correctly).

Thanks, but I'm using Win98SE and appear to have the latest driver version installed.

M-Audio 24/96

Reply #9 –
You did that experiment in 6 minutes!?

Okay, why would you want to lower the recording level? If it's to prevent the possibility of clipping (which is the only good reason I can think of) then the problem is that lowering the volume on the mixer doesn't prevent clipping. It's exactly like recording as it, then lowering the level in Cool Edit (or whatever). The clipping is still there and you have raised the noise floor.

If we're not talking about preventing clipping, then yes, lowering the volume does not cause a lot of degradation. This is how replaygain is done after all, and with proper dither (which I don't know if the M-audio mixer does) all you get is a slightly higher noise floor. For playback I would not sweat getting up to change the volume (especially if you have a cat in your lap like I do right now  ) when you could just use the digital mixer. But you should be aware that in theory lowering the volume at the amp would be better.

M-Audio 24/96

Reply #10 –
Quote
You did that experiment in 6 minutes!?

Okay, why would you want to lower the recording level? If it's to prevent the possibility of clipping (which is the only good reason I can think of) then the problem is that lowering the volume on the mixer doesn't prevent clipping. It's exactly like recording as it, then lowering the level in Cool Edit (or whatever). The clipping is still there and you have raised the noise floor.

If we're not talking about preventing clipping, then yes, lowering the volume does not cause a lot of degradation. This is how replaygain is done after all, and with proper dither (which I don't know if the M-audio mixer does) all you get is a slightly higher noise floor. For playback I would not sweat getting up to change the volume (especially if you have a cat in your lap like I do right now   ) when you could just use the digital mixer. But you should be aware that in theory lowering the volume at the amp would be better.

I didn't do the experiment, but understood what you meant.  I *did* switch around my external preamplifier so I could use the volume control rather than just the phono stage output... trouble is, using it this way adds significant *hissssss* at low levels when the volume is turned up past the detent level.  I think things were better (in terms of noise floor) with the Soundblaster PCI128 line-in and just using the phono stage!  So in a way, I wasted my money on this card unless I ever get a MIDI keyboard and/or get into using software synths & things of that nature.

As for using the mixer on playback, maybe it's negligence on my part that I failed to state I'm using headphones connected directly to the analog output of the soundcard... too lazy to go back and see if I said this or not. 

I do appreciate your input (pun semi-intended) on all this... thanks mucho.

M-Audio 24/96

Reply #11 –
Well, you could just record from the phono preamp directly and boost the volume in your editor after recording (although don't do this in 16-bit mode). If you get less hiss that way, then that's the way to do it.

A friend of mine has that card and says it won't drive his HD-580s with any bass (so he's been borrowing my headphone amp), although those are particularly difficult to drive. Keep that in mind if you do upgrade your phones.

Glad I could help...

M-Audio 24/96

Reply #12 –
Quote
A friend of mine has that card and says it won't drive his HD-580s with any bass (so he's been borrowing my headphone amp), although those are particularly difficult to drive. Keep that in mind if you do upgrade your phones.

Glad I could help...

It's definitely driving my Grado SR-60's and Denon AH-D550 headphones pretty well (at least as well as the PCI128 does), although the real test will come in if/when I watch a DVD with the sound going thru the 24/96.

M-Audio 24/96

Reply #13 –
Quote
The 24/96's H-W interface seemed pretty easy to figure out, but... WTF, no handy-dandy system tray icon?  I actually have to reload this mixer thing each time I want to use it... when was this thing designed, back in 1993?  Crap.

Drag the mixer from the Control Panel onto the desktop - one instant short-cut directly to it, ready to be double clicked whenever you need it!

(some of us like the fact that it doesn't fill your system tray with crap! and you obviously didn't suffer PCs in 1993 if you think this hardware+driver software is comparable!)

Quote
Regular software volume controls on programs like WinAMP are, of course, non-functional.  Double-crap.


No, that's a good thing. Anything which tries to use the sound card to change stuff will fail. Things with high quality DSP will work though (e.g. foobar), and the output canbe 24-bit now, so no quality loss - significant quality gain probably!

Quote
Too bad the monitor mixer only allows line-in levels to be cut and not boosted... that's a pain when trying to maximize recording levels from LP's without clipping.


If they're too high already, you have a problem - as bryant has said, reducing them won't stop clipping because it's happening before the mixer. A couple of well chosen resistors in a patch cord would do the job though.

If the levels are too low - how low are they? If they peak above -10dB, don't worry about it. If you raise the result (recorded at 24-bit resolution) by 10dB in software it'll still be better than you were getting with your SB128 at the correct level. If they're lower than this, you probably need some amplification.

Don't use the headphone or speaker output of your amp just so you can use the volume control - this is a terrible way to do it - it's always way too hissy, and one knock to the volume control could toast your nice new sound card!


Quote
Questions, comments, inputs?  Can anyone help with the DirectX thing, and tell me what software records in 24 bits?  Certainly not Cool Edit Pro v1.2.


It certainly does! I have the audiophile 2496, Win98SE, and CEP 1.2a and it works as advertised - what settings are you using? Have you had a look at the syntrillium website?

Cheers,
David.

M-Audio 24/96

Reply #14 –
Quote
Don't use the headphone or speaker output of your amp just so you can use the volume control - this is a terrible way to do it - it's always way too hissy, and one knock to the volume control could toast your nice new sound card!

(snip)

It certainly does! I have the audiophile 2496, Win98SE, and CEP 1.2a and it works as advertised - what settings are you using? Have you had a look at the syntrillium website?

Cheers,
David.

Thanks for the comments, they did give me a different perspective on things...

Just to address the two above.  I'm using the output on my preamp, but this is a "true" preamplifier (vintage 1979, high-end consumer model) and doesn't output very much even if the volume control is all the way up.  At about middle position (5) it outputs about at the level of the built-in phono stage, and lower volume settings just result in an (analog) cut in volume from line-level.  I rather like it, since it allows me to adjust channel balance in analog as well (my turntable's cartridge is a bit out of spec and favors the right channel), and perhaps boost bass or treble slightly with records that sound dull, or use a subsonic filter with records having a lot of "groove scrub" noise -- all using analog circuitry, before the signal is ever recorded.

The second point -- Cool Edit Pro 1.2a has no 24-bit recording level, at least none I found.  What it appears to have is 8-bit, 16-bit, and 32-bit float.  Maybe I have v1.2 and not v1.2a.

Anyway, I'm a Sound Forge die-hard, but can't afford to upgrade to a newer version that supports better than 16-bit recording (it does support all the rates of the card up to 96KHz).  Since 16 bits of dynamic range ought to be good enough for vinyl, I don't see a lot of point in using more than that.

M-Audio 24/96

Reply #15 –
Quote
The second point -- Cool Edit Pro 1.2a has no 24-bit recording level, at least none I found.  What it appears to have is 8-bit, 16-bit, and 32-bit float.  Maybe I have v1.2 and not v1.2a.

In cooledit you will not get the 24bit setting until you save the file. You have to start with a 32bit wave stream. The 24bit option is in the Save As dialog, save the stream as wave, and in the Options dialog, choose 24bit packed int, then you will be able to get a 24bit wave file.

Hope this help!

M-Audio 24/96

Reply #16 –
Quote
Quote
The second point -- Cool Edit Pro 1.2a has no 24-bit recording level, at least none I found.  What it appears to have is 8-bit, 16-bit, and 32-bit float.  Maybe I have v1.2 and not v1.2a.

In cooledit you will not get the 24bit setting until you save the file. You have to start with a 32bit wave stream. The 24bit option is in the Save As dialog, save the stream as wave, and in the Options dialog, choose 24bit packed int, then you will be able to get a 24bit wave file.

Hope this help!

Thanks, it does... I'm just surprised Cool Edit (Pro) won't record directly in 24-bit format with 24/96 cards.  What's happening to the 24-bit int data as it gets converted to 32-bit float while recording, I wonder?  Just represented as 32-bit floating point numbers?

I guess that's good with such things as normalizing, compressing, equalizing, etc (probably much more exact).  I just wonder, upon re-conversion to 16-bit integer... is it a true representation of the original sound?  I'd think there would have to be a lot of rounding done.  Because of that, the use of 32-bit float seems a rather "artificial" way of preserving sound quality to me (someone correct me if I'm wrong).

M-Audio 24/96

Reply #17 –
Quote
Thanks, it does... I'm just surprised Cool Edit (Pro) won't record directly in 24-bit format with 24/96 cards.  What's happening to the 24-bit int data as it gets converted to 32-bit float while recording, I wonder?  Just represented as 32-bit floating point numbers?

A 32-bit float will losslessly store any 24-bit int, and it's much better once you start manipulating because you never have to worry about clipping or underflow and you never lose any resolution (so you don't need to dither at every step). Also, modern CPUs handle 32-bit floats just as quickly as integers, and would actually be faster with floats than with packed 24-bit ints.

Storing 32-bit floats as 24-bit ints is not lossless unless the 32-bit floats came directly (without manipulation) from a 24-bit source, but the loss is trivial. The only disadvantage to using 32-bit floats internally is that it takes a little time to load and store them if you want 24-bit files externally.

M-Audio 24/96

Reply #18 –
Quote
Quote
Thanks, it does... I'm just surprised Cool Edit (Pro) won't record directly in 24-bit format with 24/96 cards.  What's happening to the 24-bit int data as it gets converted to 32-bit float while recording, I wonder?  Just represented as 32-bit floating point numbers?

A 32-bit float will losslessly store any 24-bit int, and it's much better once you start manipulating because you never have to worry about clipping or underflow and you never lose any resolution (so you don't need to dither at every step). Also, modern CPUs handle 32-bit floats just as quickly as integers, and would actually be faster with floats than with packed 24-bit ints.

Storing 32-bit floats as 24-bit ints is not lossless unless the 32-bit floats came directly (without manipulation) from a 24-bit source, but the loss is trivial. The only disadvantage to using 32-bit floats internally is that it takes a little time to load and store them if you want 24-bit files externally.

Thanks Bryant, and a final question -- suppose I were to record at 44.1KHz/32-bit float, then convert to a regular CD-quality 16-bit .wav... how about losses in that case?  For some reason, I can't shake the idea that there would have to be serious losses in that sort of conversion -- how does the software know a value like 32.50123, should it be 32 or 33?

32-bit float may be lossless as far as resolution, but I've never heard of a soundcard that could play 32-bit floating point wave files (!).  Would I be better off just recording at 16-bits if I weren't planning to do much manipulation of the file (perhaps just a single volume change)?

Thanx

M-Audio 24/96

Reply #19 –
Quote
32-bit float may be lossless as far as resolution, but I've never heard of a soundcard that could play 32-bit floating point wave files (!).  Would I be better off just recording at 16-bits if I weren't planning to do much manipulation of the file (perhaps just a single volume change)?

First of all, processing audio as 32-bit floats is not that strange. Foobar2000 used these for all its internal processing until it recently switched to 64-bit floats! At the very last step the floating number is converted to the appropriate output bitdepth using well-understood principles of dithering and noise shaping to preserve as much fidelity as possible.

In your case, the best possible method would be to convert the 24-bit audio from your card to 32-bit floats during recording, do all processing as floats, then as the absolute last step convert to 16-bit (with dither and noise shaping).

The worst thing you could do would be to convert the 24-bit audio from your card directly to 16-bit using truncation (which would probably happen if you used an audio editor that only supported 16-bits), then apply 10 dB of gain to a get a good level (which would also boost the noise and distortion introduced in the previous step).

In fairness, there is an alternative view to all this that should be presented. Someone could quite reasonably argue that the noise and distortion introduced by using a cheap soundcard and doing all these operations in 16-bits is far less (and completely swamped by) the noise and distortion present in vinyl records in the first place, and that all this discussion is simply mental masturbation. Since you are interested in restoring LPs in the first place though, I suspect that this extra attention to detail is probably worth something (if only piece of mind that you aren't making things any worse).

Edit: clarity

M-Audio 24/96

Reply #20 –
Quote
The worst thing you could do would be to convert the 24-bit audio from your card directly to 16-bit using truncation (which would probably happen if you used an audio editor that only supported 16-bits), then apply 10 dB of gain to a get a good level (which would also boost the noise and distortion introduced in the previous step).

I don't get it... a 24-bit capable card (specifically the M-Audio 24/96) *isn't* capable of 16-bit recording, and recording in 16-bit using Sound Forge will actually be recording in 24 bits with software truncation?? 

Curiouser and curiouser....   

I suppose it wouldn't be too much to ask why a 24-bit card couldn't do normal 16-bit recording like a 16-bit card could?  Sorry, it's just completely nonintuitive.  My intuition says to record in 16 bits and just leave it alone after that.

M-Audio 24/96

Reply #21 –
Quote
I suppose it wouldn't be too much to ask why a 24-bit card couldn't do normal 16-bit recording like a 16-bit card could?  Sorry, it's just completely nonintuitive.

Well, obviously it could, but I'm not at all sure that it would.

The ADC on the card generates 24-bit data and if the recording software is asking for only 16-bits then some software (or firmware) somewhere in-between has to get rid of those extra 8 bits. There might be a special routine to carefully dither and noise shape, or they might just get thrown away (truncation). Without experimenting it's impossible to know which it is (and I would guess truncation, but I'm a grumpy old pessimist).

M-Audio 24/96

Reply #22 –
Thanks Bryant.  I'm more lost in some ways than I was at the start of this conversation, but I guess I'll be changing my recording technique to use 32-bit float and then downconvert (uggh, but... ok).  Maybe I'll try recording in 16 bits too, and ABX to see if I can hear any difference.

Any experts here on the M-Audio 24/96 who could clear this up?  I'll check the manual carefully too, to see if there are any hints on 16-bit recording.

M-Audio 24/96

Reply #23 –
As the others say, the best performance way of recording is to record in 32-bit floating point, do all your processing and editing, and finally convert to to 16 bit with dithering.

It may not lead to audible improvements if you do little or no processing, but it takes little extra effort, so it's a good idea to use it always, so that you get always max possible quality.

As to recording: don't use the mixer when recording. You won't get any quality benefit in any case, and can have quality issues depending on the dynamics of the recorded signal. It's better to do all the level corrections after recording, using a high quality 32 bit float wave editor.

As to it not sounding very different from a SB128: non-crappy soundcards, such as some SB128, are actually quite good, so that's not surprising. Some people at HA use a SB128 for critical listening.

M-Audio 24/96

Reply #24 –
Oh - and no one mentioned dither to really confuse fewtch!


http://mp3decoders.mp3-tech.org/24bit.html
http://www.mtsu.edu/~dsmitche/rim420/readi...420_Dither.html


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!


For one thing, you've suggested that you're going to increase the volume in software. This means there's a significant advantage to recording 24-bit data. Let's say you're going to doube the volume. Doubling in binary is like multiplying by ten in decimal - you move everything one place to the left. With a 16-bit recording at the start, and a 16-bit recording at the end, the last bit is always going to be 0 if you multiply by 2, because there's no data below the 16th bit to put in it!

OTOH if you start with 24-bits, double the volume, and convert to 16-bits, you've actually got some useful data in the 16th bit. What's more, with dither, you're keeping some of the information from all the other bits too.

That second part falls into the "mental masturbation" category. But having less noise in the original file (24-bit vs 16-bit) is certainly useful when you're going to increase the loudness in software.

Cheers,
David.