The Official Sequential/DSI Forum

Adding drum sounds from the Tempest into Ableton without analog noise?!

Shea

You do have to be careful maxing out volume on the T though, because things will start to clip, especially with the digital sine wave, and especially when played polyphonically.

Not having these kind of issues and trust me i'm really squeezing that headroom..Just be careful with the master out..
You do have to be careful maxing out volume on the T though, because things will start to clip, especially with the digital sine wave, and especially when played polyphonically.

I'm running all voices out separately, but only the left channel of each voice for now...
Directly into my RME Fireface 400
At unity the meter in Live is reading 0.00 but it doesn't seem to clip.
I just pull the fader in Ableton down to a manageable volume and done.
Although I have 24 bits, I'm not in the habit of wasting too many of them.
If you're working so far under 0.00, you're effectively reducing your bit-rate.

idm

No, you're not. That's not how it works. :). If you record something at 24  bit and very very low in amplitude, it's not suddenly 8 bit or something like that.

@ 24 bit recording highly dynamic sounds is just the way the go because recording at lower levels (-6db or - 12db even) has no negative impact on your sound.

Recording something really loud into ableton and lowering the track volume is the opposite of how you would record something as clean as possible. I guess RME has some incredible hi-gain inputs so that's what is saving your sound. But it's certainly not how you cleanly record sounds. You are just smashing hi gain sound through the inputs and then lower the gain in software. There's no good reason to do that.

I use proper gain staging so I can leave my level faders in my DAW at their default value. Recording the signal so it hits - 12db max, works perfectly fine, and if the signal then has a peak, it has 12 db of headroom for it not to clip. Which is just convenient.

It's on the Tempest where you have to worry about the levels being high enough. You can leave a lot of headroom at the AD stage if recording in 24bit as others have said.

I never have a problem with noise (other than what one would expect in the analog domain) unless using loads of compression or distortion but, again, that is to be expected.

At the risk of repeating what others have said.

- Make sure Tempest mixer levels are maxxed out.

- Make sure 'amount' on 'amp env' is set to max (127), or at least as high as you can get away with if you're doing samples at different velocities.

- If the sound is mono record one channel and pan hard left or right. Don't set the mixer panned centre as that will decrease gain on each channel (I'm pretty sure the T applies some kind of pan law from memory).

- Use the individual voice outs, this could be my imagination but they always seem hotter and clearer to me than the main outs.
Noise, Noodles and Doodles: http://bit.ly/mrjonesthebutcher

I don't know where you've heard these things but they're not correct..At least in my audio engineer school (SAE) we've been instructed to record signals around 0db (of course not clipping).Your individual track levels must be recorded around 0db and do your mixing so your master hits around -6db.Then stick your compressors/mastering chains to boost it up to 0db..24bit recordings are always better cause they lift up the dynamic range but don't get fooled if you're recording at lower levels and want to go to 0db with compression etc.you're also lifting up that noise floor.

Cheers!
No, you're not. That's not how it works. :). If you record something at 24  bit and very very low in amplitude, it's not suddenly 8 bit or something like that.

@ 24 bit recording highly dynamic sounds is just the way the go because recording at lower levels (-6db or - 12db even) has no negative impact on your sound.

Recording something really loud into ableton and lowering the track volume is the opposite of how you would record something as clean as possible. I guess RME has some incredible hi-gain inputs so that's what is saving your sound. But it's certainly not how you cleanly record sounds. You are just smashing hi gain sound through the inputs and then lower the gain in software. There's no good reason to do that.

I use proper gain staging so I can leave my level faders in my DAW at their default value. Recording the signal so it hits - 12db max, works perfectly fine, and if the signal then has a peak, it has 12 db of headroom for it not to clip. Which is just convenient.
« Last Edit: June 30, 2017, 02:35:37 AM by Yorgos Arabatzis »

No, you're not. That's not how it works. :). If you record something at 24  bit and very very low in amplitude, it's not suddenly 8 bit or something like that.

@ 24 bit recording highly dynamic sounds is just the way the go because recording at lower levels (-6db or - 12db even) has no negative impact on your sound.

Recording something really loud into ableton and lowering the track volume is the opposite of how you would record something as clean as possible. I guess RME has some incredible hi-gain inputs so that's what is saving your sound. But it's certainly not how you cleanly record sounds. You are just smashing hi gain sound through the inputs and then lower the gain in software. There's no good reason to do that.

I use proper gain staging so I can leave my level faders in my DAW at their default value. Recording the signal so it hits - 12db max, works perfectly fine, and if the signal then has a peak, it has 12 db of headroom for it not to clip. Which is just convenient.

You're wrong.
If you're not using the 24 bits of amplitude resolution, then you are effectively lowering your bit rate. Record as hot as you can, without clipping, then lower the volume. Recording at 12 dB under, then boosting the volume of the file, is effectively increasing your noise floor by 12 dB. The way you describe is definitely NOT the way to get the clearest recording. We're talking about digital audio files, which are 'clearest' at their loudest point. Analog however would be cleaner around the -12dB mark.

and I'm not saying that 'all of a sudden your file is 8-bit'...
but you WOULD actually be only utilising 8-bits of your available 24.

and I'm not saying that 'all of a sudden your file is 8-bit'...
but you WOULD actually be only utilising 8-bits of your available 24.

Not true. You would have to be recording at ridiculously low levels to only be utilising 8 bits, something like -100db.

-12b in a 24bit system is nothing. Generally your tracks will have to be reduced by at least this anyway so as to leave a decent amount of dynamic range on the master bus.

So, whilst it is technically 'best' to record as hot as possible, there's no sense being paranoid about it. It's far worse to find your recording's been clipping accidentally. DO get the output from the Tempest as hot as possible, DON'T sweat it too much about getting the input level too close to 0db on your audio interface. Unless your audio interface has ridiculously high noise floor it will make no difference.

Here's a good article about bit depth and dynamic range and why you shouldn't sweat it too much:

http://www.mojo-audio.com/blog/the-24bit-delusion/

I've done tests in the digital domain using Logic (which admittedly is a 32bit system) reducing track levels by well below realistic levels (I can't remember exactly what I did but it was stupid amounts), increasing gain back to the original level and then phase-reversing with the original signal to check noise floor and it was well below what would be perceptible at anything like normal listening levels.

Noise, Noodles and Doodles: http://bit.ly/mrjonesthebutcher

idm

^ this.

Yorgos Arabatzis :

I don't know where you've heard these things but they're not correct..At least in my audio engineer school (SAE) we've been instructed to record signals around 0db (of course not clipping).Your individual track levels must be recorded around 0db and do your mixing so your master hits around -6db..

^


This right here is why my mentor and so many others stopped taking on students from schools. Very sad they do/still teach kids this way. Audio learned from a book.

The reason people probably still have this paranoia about recording as close to 0db as possible is that they come from the 'old school' days of tape where the medium you were recording on to had an inherent noise floor that was really quite significant so you needed to get as far above that as you could.

There is no such issue with 24bit digital recording. Yes, there will be some in the analog components of your audio interface but it would really have to be a pretty crappy audio interface for 12db to make any difference.

The noise floor in the Tempest though is significant, so that's why you want to boost the signal out of the T as high as possible before it goes anywhere else!
Noise, Noodles and Doodles: http://bit.ly/mrjonesthebutcher

In fact we've learned at SAE to edit first in tape and then we moved to DAWs so i guess i'm old school ;)
The reason people probably still have this paranoia about recording as close to 0db as possible is that they come from the 'old school' days of tape

Yorgos Arabatzis :

I don't know where you've heard these things but they're not correct..At least in my audio engineer school (SAE) we've been instructed to record signals around 0db (of course not clipping).Your individual track levels must be recorded around 0db and do your mixing so your master hits around -6db..

^


This right here is why my mentor and so many others stopped taking on students from schools. Very sad they do/still teach kids this way. Audio learned from a book.

Every audio interface + input signal path + ADC will be different as is each design; clearly, it's good and recommended to get a "decent" amount of dynamic range for individual tracks within your DAW, though there are some situations where the analogue input circuitry of your audio interface might in fact overload (in some cases, intentionally) or soft-clip before 0dB FS is reached.

You have to experiment to find the best possible balance, as with anything audio-related.
« Last Edit: June 30, 2017, 06:07:16 AM by DavidDever »
Sequential / DSI stuff: Prophet-6 Keyboard with Yorick Tech LFE, Prophet 12 Keyboard, Mono Evolver Keyboard, Split-Eight, Six-Trak, Prophet 2000

getting taken out of context here. especially with this "8-bit" shenanigans.
whatever, all I'm saying is, to make the most of your bits, friggon USE them.
if everything you do only makes a tiny bit of difference, in the end - it all adds up.

idm

getting taken out of context here. especially with this "8-bit" shenanigans.
whatever, all I'm saying is, to make the most of your bits, friggon USE them.
if everything you do only makes a tiny bit of difference, in the end - it all adds up.

Again... This is not how things work. Please read the article a couple of posts above, it explains it very well.

For you to enjoy every single "bit" at 24bit, you'd have to play your recordings, that you smashed as loud as possible into your DAW, back at eardeafening level. It makes no sense.

 For recording purposes it is useful to record at 24bit because you can record at lower levels without hearing noise which you would get when you'd record at low levels @ 16 bit. This doesn't mean you lose bits. It's a really strange way to put it, and it's just plain wrong. Nobody is recording @ - 48db here, you understand? :)

Recording something at 8bits will yield completely different results than recording at incredible low levels (and by that I mean - 60db or less). So recording at - 12db is not a problem at all and has 0,0 influence on your sound, the chance of clipping is reduced though, which is the whole point.

SAE can say what they want (a very very long time ago though... I'm sure they don't teach it like that anymore). Mixing  at 0db is a must on analog mixing tables, not in the digital world. You're better of keeping your channels at - 6db or even - 12 db, as it has zero influence on your sound, other than giving you way more headroom at the master channel.
I studied music production as well,  recorded many bands, and mix and master electronic music pretty regularly for other producers. So I'm not  making this stuff up on the spot ;)

Hope I don't sound too annoyed haha. Don't mean it ike that ;)
« Last Edit: June 30, 2017, 09:08:50 AM by idm »

So recording at - 12db is not a problem at all and has 0,0 influence on your sound, the chance of clipping is reduced though, which is the whole point.

Agreed - it may be the only safe way to avoid clipping in both the analogue and digital domains. Even if you believed that you "lose" a relatively few bits of resolution in reduced dynamic range, you'd still be getting a 20-bit signal, which (with higher sample rates) is plenty fine enough for multitrack audio recording. (In practice, you don't actually lose any bits.)
« Last Edit: June 30, 2017, 11:16:08 AM by DavidDever »
Sequential / DSI stuff: Prophet-6 Keyboard with Yorick Tech LFE, Prophet 12 Keyboard, Mono Evolver Keyboard, Split-Eight, Six-Trak, Prophet 2000

again, you're taking what i'm saying, and arguing out of context