Soft synth D/A question

For computer based music makers. Discussions about plug-ins and stand alone computer synth gear.
KrisH
Newbie
Newbie
Posts: 33
Joined: Mon Apr 19, 2010 3:43 pm
Real name: Kris
Gear: Roland D-50
Roland HS-60
Ensoniq SD-1
Yamaha DX11
Kawai K5
Akai Miniak
Casio DG-20
CME VX5
StudioLogic MP-113
Gateway 4026GZ
Acer Aspire 7720
Band: Hofenakus;Hölt
Contact:

Soft synth D/A question

Post by KrisH » Mon Feb 07, 2011 4:09 am

I have several A/D/A converters, two used primarily for connecting my laptop to my amplifiers. I understand how bit depth and sampling rate can affect recording quality, but I'm confused as to how it might affect the sound quaity of the soft synths. For performance, is sample rate more important than bit depth, or vice-versa? In other words, which would be preferable: 16 bit 48 mHz or 24 bit 44.1 mHz. Or doesn't this matter in sound quality when converting digital to analog? I've tried to research this and just get confused. What other factors should I consider?

Ry-Fi
Supporting Member!
Supporting Member!
Posts: 203
Joined: Thu Jan 31, 2008 9:40 am
Gear: Ion, MKS-30, MKS-50, SQ-1+ 32-voice, TX81Z
Location: Regina, SK
Contact:

Re: Soft synth D/A question

Post by Ry-Fi » Mon Feb 07, 2011 4:54 am

In very short, without addressing a lot of the surrounding issues and debates:

Bit depth = dynamic range ("resolution"), and sample rate = bandwidth.

As long as your sample rate is at least 44.1 kHz, the range of human hearing (20 Hz - 20 kHz) is preserved and you needn't go higher. Of course, there are additional issues with regards to the reconstruction filter in the DAC. While higher sampling rates allow for greater than 20 kHz content, the major advantage is allowing for a reconstruction filter with more accurate response in the extreme highs (near 20 kHz).

A bit depth higher than 16 bits is desired, as 16 bits allow for a maximum dynamic range of ~96 dB. Since human hearing has a dynamic range of 130 dB+, this clearly isn't quite enough for sonic perfection. 24 bits is far more desirable, as this allows for dynamic ranges of up to ~144 dB. Having said that, no converters and their surrounding output circuits are even capable of achieving 144 dB dynamic ranges/signal-to-noise ratios. At most, you can get about 21 bits worth of resolution out of modern pro audio electronics. This, however, is still far preferable to 16 bits.

I say go with 24 bit @ 44.1 kHz, over 16 bit @ any other sample rate. Of course, that's only a good idea *in principle*. Your DAC could conceivably sound better/worse at different bit depths and sample rates...

Ashe37
Moderator
Moderator
Posts: 3790
Joined: Mon Jan 19, 2009 3:43 pm
Real name: Unpronounceable
Gear: Ensoniq SD-1/32,SQR,VFX,ESQm
Virus Indigo, M3-61 , MS2000BR, Volca Bass
Emu XL-7, Matrix 6r
TG-33, K3m, Blofeld, Micron, Mopho, BS II, JV-1080
Band: Eridani V
Location: Central VA

Re: Soft synth D/A question

Post by Ashe37 » Mon Feb 07, 2011 5:22 am

The only reason to use 48 KHz, officially, is for video and film work. most sound systems won't reproduce anything above 20 KHz anyway.....

KrisH
Newbie
Newbie
Posts: 33
Joined: Mon Apr 19, 2010 3:43 pm
Real name: Kris
Gear: Roland D-50
Roland HS-60
Ensoniq SD-1
Yamaha DX11
Kawai K5
Akai Miniak
Casio DG-20
CME VX5
StudioLogic MP-113
Gateway 4026GZ
Acer Aspire 7720
Band: Hofenakus;Hölt
Contact:

Re: Soft synth D/A question

Post by KrisH » Mon Feb 07, 2011 1:53 pm

Thanks for the responses and comments. My thinking is that the higher sample rate might work to reduce aliasing, but I suppose that's probably an issue handled by the plugins themselves. I've noticed that the lower bit depth seems to use less CPU resources. I don't use this rig for recording (have a 24 bit 48 kHz converter for that), so is more than a 96 db range practically usable?

User avatar
meatballfulton
Moderator
Moderator
Posts: 5774
Joined: Wed Apr 13, 2005 9:29 pm
Gear: Live 9, Logic Pro X

Re: Soft synth D/A question

Post by meatballfulton » Mon Feb 07, 2011 2:11 pm

KrisH wrote:I don't use this rig for recording (have a 24 bit 48 kHz converter for that), so is more than a 96 db range practically usable?
No, it's not usable. Unless you are playing in a completely silent environment the noise floor of the space you are in determines your usable dynamic range.

For reference, vinyl LPs and cassette tapes only have a dynamic range of about 50 dB.
I listened to Hatfield and the North at Rainbow. They were very wonderful and they made my heart a prisoner.

User avatar
madtheory
Supporting Member!
Supporting Member!
Posts: 5155
Joined: Mon May 01, 2006 12:45 pm
Real name: Tomas Mulcahy
Gear: Flangebeast Mk1, Plonkotron, Morovdis Arpeggiator, Maplin My First EQ, Jeff Wayne Thunderchild rack, Thermostat, Buck Owens' Moog.
Band: Minim
Location: Cork, Ireland
Contact:

Re: Soft synth D/A question

Post by madtheory » Mon Feb 07, 2011 3:51 pm

I disagree, 16 bit is usable. To overcome the high noise floor in your environment, either reduce the sources of noise, or turn up the loudspeakers! Within reason of course.

If you want to compose with soft synths and running 16 bit is convenient, that's totally sensible. You can always change the session to 24 bit later (most DAWS allow that if you haven't rendered/ frozen tracks). Alternatively, I think you could change to a dual core i5 processor or better, those things have plenty of power for audio and music work.

Ry-Fi, I don't think it's sensible to equate dBSPL with the dynamic range at 24 bit. While it is true that we can hear from 0dBSPL to about 130dBSPL, reproducing that in an audio system is very difficult, unrealistic and dangerous. Above 85dBSPL it is impossible to avoid permanent damage even for short duration exposure.

So why do we use 24 bit? For headroom, where you can set the peak level at -18dBFS thereby avoiding clipping entirely, staying well above the noise floor with the bonus of not having to watch levels like a hawk. It also allows improved resolution compared to 16 bit when processing the sampled audio signal, notwithstanding the fact that most plugins process at 32 bit FP or greater.

User avatar
meatballfulton
Moderator
Moderator
Posts: 5774
Joined: Wed Apr 13, 2005 9:29 pm
Gear: Live 9, Logic Pro X

Re: Soft synth D/A question

Post by meatballfulton » Mon Feb 07, 2011 7:29 pm

madtheory wrote:I disagree, 16 bit is usable
OP asked if dynamic range greater than 96 dB is usable for live playing, I wrote no so we in fact agree 8-)
I listened to Hatfield and the North at Rainbow. They were very wonderful and they made my heart a prisoner.

User avatar
madtheory
Supporting Member!
Supporting Member!
Posts: 5155
Joined: Mon May 01, 2006 12:45 pm
Real name: Tomas Mulcahy
Gear: Flangebeast Mk1, Plonkotron, Morovdis Arpeggiator, Maplin My First EQ, Jeff Wayne Thunderchild rack, Thermostat, Buck Owens' Moog.
Band: Minim
Location: Cork, Ireland
Contact:

Re: Soft synth D/A question

Post by madtheory » Mon Feb 07, 2011 8:10 pm

No we don't. As I said, it's easy to overcome the noise floor. Also in this instance, there's a reason why dynamic range expressed as a ratio in dB is not the same thing as environmental noise measured in dBSPL.

Ry-Fi
Supporting Member!
Supporting Member!
Posts: 203
Joined: Thu Jan 31, 2008 9:40 am
Gear: Ion, MKS-30, MKS-50, SQ-1+ 32-voice, TX81Z
Location: Regina, SK
Contact:

Re: Soft synth D/A question

Post by Ry-Fi » Mon Feb 07, 2011 10:38 pm

madtheory wrote:I disagree, 16 bit is usable. To overcome the high noise floor in your environment, either reduce the sources of noise, or turn up the loudspeakers! Within reason of course.

If you want to compose with soft synths and running 16 bit is convenient, that's totally sensible. You can always change the session to 24 bit later (most DAWS allow that if you haven't rendered/ frozen tracks). Alternatively, I think you could change to a dual core i5 processor or better, those things have plenty of power for audio and music work.

Ry-Fi, I don't think it's sensible to equate dBSPL with the dynamic range at 24 bit. While it is true that we can hear from 0dBSPL to about 130dBSPL, reproducing that in an audio system is very difficult, unrealistic and dangerous. Above 85dBSPL it is impossible to avoid permanent damage even for short duration exposure.
I don't disagree with you at all, as you're of course right on all points. I kept my answer short, and these are the points I didn't bring up. Of course, 130 dB is dangerous, NO ONE ever listens to music on an excellent stereo in an environment with an ultra low noise floor, most music is listened to on shitty mp3 players, etc...

User avatar
madtheory
Supporting Member!
Supporting Member!
Posts: 5155
Joined: Mon May 01, 2006 12:45 pm
Real name: Tomas Mulcahy
Gear: Flangebeast Mk1, Plonkotron, Morovdis Arpeggiator, Maplin My First EQ, Jeff Wayne Thunderchild rack, Thermostat, Buck Owens' Moog.
Band: Minim
Location: Cork, Ireland
Contact:

Re: Soft synth D/A question

Post by madtheory » Mon Feb 07, 2011 10:52 pm

I understand, sometimes it is difficult to decide what to leave out when explaining technical issues such as this. I encounter this problem every day in my work, it's a challenge but it's a great way to learn. I call them white lies. :)

Another point I think is that if you're using a 16 bit converter to reproduce all your sounds, it will be OK because you have complete control over the dynamics, and most PA systems have limiting because of ambient noise, SPL regulations and to protect the speakers.

astroidmist
Junior Member
Junior Member
Posts: 216
Joined: Mon Aug 16, 2010 7:55 am

Re: Soft synth D/A question

Post by astroidmist » Sat Feb 12, 2011 4:21 am

16-bit can sometimes sound a bit grainy if you're not careful. But filtering can sometimes get rid of that graininess after the fact. I'm talking about quantization noise in this case.

24-bit doesn't have that issue and also allows for more headroom during recording which is really really important and is reason enough to go with 24 instead of 16.

But I agree 44.1 kHz (not mHz) is reasonable.

User avatar
Shreddie
Junior Member
Junior Member
Posts: 232
Joined: Tue May 12, 2009 7:50 pm
Gear: Alesis Andromeda, Fusion 8HD: ART SGX Nightbass SE: Kawai K5000s: Yamaha A3000 V2, EX7, FS1R, PSS-480 (Don't laugh, it rocks!) and 4 planks.
Location: UK

Re: Soft synth D/A question

Post by Shreddie » Sun Feb 13, 2011 2:00 am

Ry-Fi wrote:A bit depth higher than 16 bits is desired, as 16 bits allow for a maximum dynamic range of ~96 dB. Since human hearing has a dynamic range of 130 dB+, this clearly isn't quite enough for sonic perfection. 24 bits is far more desirable, as this allows for dynamic ranges of up to ~144 dB. Having said that, no converters and their surrounding output circuits are even capable of achieving 144 dB dynamic ranges/signal-to-noise ratios. At most, you can get about 21 bits worth of resolution out of modern pro audio electronics. This, however, is still far preferable to 16 bits.
The sensitivity of human hearing has nothing (or next to nothing) to do with it. If you were listening at levels where there were 130dB peaks, you would be in severe pain, listening at the 96dB level of 16 bit systems is not reccommended either as that would damage your hearing after prolonged listening.

There are advantges to 24 bit though, a proportinally lower noise floor and a higher dynamic range. Processing/summing etc is also improved at 24 bit...

However, 16 bit is perfectly adequate in 99% of situations (some of what are considered the best ever records were recorded/mixed at 16 bit or less) and can have a distinct advantage in samplers... In samplers, it reduces the amount of storage space required by a third over 24 bit and reduces processing overhead too which can increase available polyphony. Also, in a busy mix, no-one is going to notice the subtle nuances and lower noise floor of a 24 bit recorded piano over the same piano recorded at 16 bit... Unless they have Superman's super hearing skillz.

However, with todays modern soft samplers running on modern powerful computers, most users will rarely run into any polyphony problems so you may as well sample at 24 bit... And as a semi-pro samplist/sound designer, that's what I do... I only convert to 16 bit for hardware formats.

Ry-Fi
Supporting Member!
Supporting Member!
Posts: 203
Joined: Thu Jan 31, 2008 9:40 am
Gear: Ion, MKS-30, MKS-50, SQ-1+ 32-voice, TX81Z
Location: Regina, SK
Contact:

Re: Soft synth D/A question

Post by Ry-Fi » Sun Feb 13, 2011 2:51 am

Shreddie wrote: The sensitivity of human hearing has nothing (or next to nothing) to do with it.
<snip>
There are advantges to 24 bit though, a proportinally lower noise floor and a higher dynamic range. Processing/summing etc is also improved at 24 bit...
These 2 statements quite obviously contradict.

The higher dynamic range / noise floor of 24-bit vs 16-bit IS AN issue, because after mixing countless channels of 16-bit audio, you've reduced your dynamic range and pushed your accumulated noise up into the range where the errors ARE well within the range of audibility, if listening happens at high enough volumes (the only way such losses of fidelity could be apparent). By mixing numerous channels of 24-bit (or let's say >20-bit) audio, the resultant dynamic range is still greater than that which we can discern and the noise floor is kept out of range of audibility, even at extreme listening levels (which is the only time the audibility of these factors is possible).

It's all based around the potential of human hearing. If we acknowledge that the better noise floor and higher dynamic ranges of > 16-bit samples are worthwhile, it's because it IS pertinent to our ability to hear them in the best case scenarios. If we could never hear them, then it would be fallacious to argue that the improved performance of 24-bit audio has any value...

I certainly wasn't interested in a debate. I'm a professional recording engineer and I know what I'm talking about and I don't need to prove it to anybody. I merely presented an extremely simple breakdown for the OP, while acknowledging that I certainly wasn't attempting to dissect the whole issue. Of course, my initial post was in the vain of presenting the simple facts, while ignoring the context that may have been present. Yes, 16-bit audio is only 2/3 the size of 24-bit audio, and thus potentially much more practical in many situations. Also the advanced capabilities of 24-bit audio may be completely wasted on synthesized sounds, in situations where you're not able to capture >16-bit fidelity to begin with, in situations of few track counts, etc....

If my choice of wording or way I presented the facts was less than ideal, I apologize. (I think I might just quit trying to be informational on the Internet, because unless you say EVERYTHING there is to say about an issue, someone's going to come along and point out something that you've missed and try to "outsmart" you. Oy.......)

I just don't agree that saying "human hearing has nothing (or next to nothing) to do with it" is correct. That's what it's ALL about! But whatever, I'm done with this thread.

User avatar
Shreddie
Junior Member
Junior Member
Posts: 232
Joined: Tue May 12, 2009 7:50 pm
Gear: Alesis Andromeda, Fusion 8HD: ART SGX Nightbass SE: Kawai K5000s: Yamaha A3000 V2, EX7, FS1R, PSS-480 (Don't laugh, it rocks!) and 4 planks.
Location: UK

Re: Soft synth D/A question

Post by Shreddie » Sun Feb 13, 2011 6:23 pm

Ry-Fi wrote:
Shreddie wrote: The sensitivity of human hearing has nothing (or next to nothing) to do with it.
<snip>
There are advantges to 24 bit though, a proportinally lower noise floor and a higher dynamic range. Processing/summing etc is also improved at 24 bit...
These 2 statements quite obviously contradict.
They're perhaps a little unclear, I concede that... But they only really contradict when taken out of context which is what you appear to have done. My first statement was with regard to the dynamic level 24 bit systems, the volume level needed for the human ear to discern that and the probelms associated with that, whereas the second statement was to do with processing within the digital domain, they're two different (if very loosely related) things... And it was those differences that I was trying to clarify for the OP.

You agreed earlier with Madtheory on the subject of very high listening volumes being dangerous and we both agree that processing/summing (and indeed your mention of accumulated noise which I hinted at when I mentioned summing) at 24 bit is advantageous so I don't understand why you seem to have taken issue with my post.

User avatar
madtheory
Supporting Member!
Supporting Member!
Posts: 5155
Joined: Mon May 01, 2006 12:45 pm
Real name: Tomas Mulcahy
Gear: Flangebeast Mk1, Plonkotron, Morovdis Arpeggiator, Maplin My First EQ, Jeff Wayne Thunderchild rack, Thermostat, Buck Owens' Moog.
Band: Minim
Location: Cork, Ireland
Contact:

Re: Soft synth D/A question

Post by madtheory » Sun Feb 13, 2011 7:27 pm

I think the OP is using softsynths, so 16 bit is fine.

Furthermore, for most of us without access to a properly sound proofed room, the ambient noise in the recording space will always be above what 16 bit can do.

Furthermore, in reality the noise floor in each overdub will be masked by the musical content of the entire mix. The mistake is in thinking that the adding is linear. This is why for a long time the best analogue tape recorder with a noise floor around -80dB was a more than acceptable medium for most purposes.

Post Reply