|
Mixing help wanted (pg. 2)
|
View this Thread in Original format
| Looney4Clooney |
| quote: | sound volume is logarithmic or something. Wikipedia is your friend.
6 channels is a bit more difficult to calculate
2 same sounds = +3dB
4 same sounds = +6dB
8 same sounds = +9dB
etc
each doubling is 3dB (8 is 2*2*2 = 3 doublings -> 3*3 = 9dB.
|
daw metering mimic vu in that they see audio and treat their representation as voltages so doubling woukd show 6dB. 3db is a doubking in power |
|
|
| DJ RANN |
| quote: | Originally posted by clay
no its all pretty much common sense for me now after i understood that 3dB is doubling of sound-pressure/volume :)
suddenly can i calculate how big amp i need for my speakers out from max effect and sensitivity and also from that calculate max spl too.
the magic is gone. |
Sorry to mind you but what size amp you need is actually completely down to the specific wattage draw of your speakers in relation to impedance, not the other way around :p
As for not worrying about noise floor or levels in the digital realm, you actually should.
WhY/ Several reasons:
1, If you're using any audio tracks or samples, they will have inherent noise recorded at a given gain, meaning if you have to pull those down to match the very low level of synths you have, then when you jack the whole thing up again to get a good master level, you're raising the relative noise floor. Not good.
2, Bad habit to get in to. If you mix without attention to proper gain staging, and then one day you have to work with a real instrument or vocalist or god forbid in a real studio, you will be absolutely ed.
3, Proper gain staging is what all DAW, software and audio technology is based one, meaning that certain functions are better aligned to having levels at the right stages at each part of the chain. For instance, when you start getting in to complex routing and grouping for say compression, you levels being at the assumed standards mean you're not running into headroom or staging issues later on in the chain.
Finally, I have actually been in a session with a big name engineer where we used up 256 physical tracks which was the max for that bus/sends/track configuration and to go beyind that dropped the desk from 32bit float to 24bit and killed the headroom.
There was literally a two hour heated conversation, mid session, between the 1st engineer and the inhouse lead engineer about how to proceed as there would be a difference in noisefloor and headroom compared to the other cues.
The result was that the master had to be lowered which is the cardinal sin of all pro mix engineers. To this day, that all involved will scowl at you if you even mention it. |
|
|
| clay |
i believe 3dB is SPL which may refer to power yes. I didnt really know mixers showed differently? but yeah dBU makes sense that its voltage lol. never thought about it. is it this way in digital world too? it really just says dB so who knows?
edit: wait a en second it only shows 0-127 lol. i need to look at a different mixer. 1sec.
edit2: my mixer (Reason) has option of selecting master between VU, Peak, PPM, VU+Peak, PPM+Peak. Who the knows what it all means, but the individual channels only shows dB - should i just assume its VU?. |
|
|
| Looney4Clooney |
| quote: | Originally posted by DJ RANN
Finally, I have actually been in a session with a big name engineer where we used up 256 physical tracks which was the max for that bus/sends/track configuration and to go beyind that dropped the desk from 32bit float to 24bit and killed the headroom.
|
most guys just use 24 bit. 32 bit float has the same precision and only introduces rounding errors so the pros that get their levels right really don't need the 8 bit mantissa. 24 bit = 32 bit float. Its just that using 32 float is like mixing in diapers. |
|
|
| clay |
| i thought the last (first?) 8 bit mantisa was only to be able to adjust master volume without affecting the actual masters dynamic? like instead of dynamic range of 0-127 you have 0,0 to 12,7 or 0,00 to 1,27? same resolution but moved in different "areas of decimals?"? i might be completely wrong but this is why you would want a 24bit dac even if you want to play only 16bit music, to be able to adjust volume digitally without destroying dynamics? or am i totally wrong? this is also why you need a 32bit dac if you want to play back 24bit music right? to be able to adjust volume digitally? (in terms of pure playback im talking about, not making music). |
|
|
| Looney4Clooney |
essentially. the output is always 24 bit. 32 bit float is just 24 bit but adding 8 bits to extend the dynamic range in a digital system. When you scale down, there will be rounding errors , there is no real advantage to using it unless you are the type of engineer that doesn't really caer about levels which is alot of people but the pros, they learned on real consoles. they don't need that safeguard. It is a bit of ego thing as well but really, there is only degradation which you probably will never hear so for normal people doing loud music, 32 float is a good idea.
if you are an assistant using a digital mixing workstation and your staging is ed but because you are using 32 f, the sound isn't really compromised, the engineer will still look at you and just shake his head because thats how they are. They hate the new breed of mixers that don't take the time to learn to do things right. 32 float exists for those that don't know how to mix properly. Its handy but there is nothing gained unless you have bad gain staging. |
|
|
| clay |
| extend isnt really the right term is it? you sort of just move it to a different scale with those last 8 bits? like if you want to mix on a low volume you dont really need a analog volume knob anymore when using 32bit because the last 8 bit allows you to adjust the master without loosing any dynamics? |
|
|
| Looney4Clooney |
| the precision is the same. I used extend as technically, that is what it does perhaps scale is a better word , but i think you get the picture. But 32f isn't an audio format. 32f = 24 bit |
|
|
| clay |
exactly. but it makes sense to get a 32bit soundcard for being able to adjust volume digitally even when playing 24bit music, as i see it. while adjusting volume on a 24bit dac will either ruin the monitored dynamics (not the project), or you will have to play back on 16bit?
edit: are you even able to set up true 24bit monitoring on your 24bit soundcard via core audio (to be able to adjust volume) or will it "freeze" at max volume so that all monitoring volume change will either have to be done inside the sequencer (in 32bit realm) or with an analog volume knob? man ive never thought about this before. or will the OS upscale the 24bit signal to 32bit (or maybe even 64bit now?) to adjust volume digitally then downscale the output again to the soundcard?
edit: what im really wondering is this. Is it stupid of me to have my windows volume on 50% while monitoring? Will I loose dynamics while monitoring (I realize the render wont be affected lol) at 50% volume using a 24bit dac?. Should i instead put the win vol on 100% and rather adjust monitoring level either inside the sequencer or analogly on my monitors (or with my analog volume knob which currently isnt hooked up). When I was using MAC the OS volume was disabled for some reason (always max) when using my 24bit dac (echo audiofire), and i wonder if this is the Reason for this. |
|
|
| Looney4Clooney |
32 bit dacs don't exist practically speaking when dealing with interfaces. I mean they might but you are talking about a dedicated DAC , not an interface and it will cost a lot and ya, just forget about it.
in terms of adjusting volume. Don't. You should calibrate your setting so you have 2 settings. 1 for normal , 1 for quiet. Or if you are doing dance, 1 for loud but ya. You should not be adjusting your interface volume. Calibrate that, leave it alone. \
Rann has written so much about this . He has reposted the link 30 times at least. |
|
|
| DJ RANN |
| quote: | Originally posted by Looney4Clooney
most guys just use 24 bit. 32 bit float has the same precision and only introduces rounding errors so the pros that get their levels right really don't need the 8 bit mantissa. 24 bit = 32 bit float. Its just that using 32 float is like mixing in diapers. |
But 24bit isn't actually 24bit, it's more like 18bit due to the fact you can't use all the bits, and nearly all 24bit DAC's can't actually deliver all 24bits in terms of usuage. Then you realize 18bit is virtually indistinguishable from 16bit but you will hear the difference between 16(or 18bit) versus a 32bit float as that is far more resolution in terms of dynamic range.
The 32-bit float has a 1-bit sign and a 23-bit mantissa, which is a total of 24b dynamic range from smallest to largest representable values. But it also has an 8-bit exponent;
In essence you can raise or lower gain by 100dB, print it, and then you can raise or lower the gain by another 100dB and you get back what you started with: no loss. You simply can't do that with a 24bit max format.
Don't forget, anything that is going to be be "re-mixed" and by that I mean a mixed stems going to at least one more stage of mixing, like the dub, that extra headroom is vital, especially when you could have literally hundreds of recorded tracks which all have inherent noise printed to them at a set floor. |
|
|
| Looney4Clooney |
forgot about the sign bit but ya, its 1 for sign -/+ 23 bit for the precision and then 8 bit for the scaling (mantissa). The point is that the added bits don't add precision.
ah i'm mixing of terms, lets just use significand
1 bit sign
23 significand
8 exponent
was conflating mantissa with the exponent. Ah and that is what you said. Ok. well that was redundant but thanks for clearing that up.
The main point is that the extra headroom isn't real. And when you start adding scaled values and then scale down, you get rounding ie degradation. There is no real reason to use it unless you are using bad gain staging. It adds nothing. It can technically make it worse although you probably won't ever notice.
i would say it is handy for those that don't give a but there isn't really anything gained. |
|
|
|
|
