Setting recording levels

September 10, 2010 4:18 PM

I could use some advice on setting recording levels for decent and consistent playback volume.

So my issue is that I do not have a set way of setting levels when recording so the end result volume varies. When it is obvious the levels are too high or low in the software (really old Cool Edit Pro 1.0) I have adjusted with amplify or normalize which gets me in the general range. Some of the songs I have recorded and then heard on other machines are too quiet or too loud. I'm looking for some basic guidelines of where to pay attention to to get proper and consistent final piece volume levels so that other people are not startled or straining to hear.

My setup is circuit bent toys --> Yamaha KM802 mixer --> desktop (winXP) line in. The toys do not have volume controls line level out. Mixer inputs I do not take above 4 (of 10) and the mixer outputs stay at 5 (of 10). The line in level is standard xp with 7 tick marks and I have that set with the slider between first and second ticks. The output of the desktop (again 7 ticks) I have the slider set between ticks 5 and 6. This goes to the line in aux input of a small bookshelf stereo (old Philips Magnavox) and that volume is set to 12 (of 40) which is a good volume (with and w/o headphones) for cds played on the desktop.

Is there an obvious spot I should focus on the levels? I know if I adjust one spot I'll need to compensate in others just not sure which is the best place. I had thought mid levels on the mixer output was best focus.

Thanks for any advice.
posted by sailormouth (12 comments total)

The levels are important everywhere.

You should be recording as loud as you can without clipping at all stages. If you do not have good metering then it's hard and you have to use your ears. The only volume that actually matters in terms of the final listener is the master volume, which should be peaking just below 0dB.

The other thing that makes a lot of difference in perceived loudness is compression/limiting, but it doesn't sound like you are using any of that.

As a rough guide, start with everything set to 7/10 and listen carefully for distortion at the loudest parts. If you hear it or can see it on the waveform, reduce the level until it disappears. Overall I suspect your levels are somewhat low.
posted by unSane at 8:18 PM on September 10, 2010


Everything unSane said. But I think some metering closer to the input of your PC would help a lot. Yes, ultimately you have to use your ears as to what sounds good but the response characteristics of different equipment (and different combinations) will make the process kind of arduous.

You might consider either getting another mixer, or using a preamp. As unSane suggests, an inline limiter will help too, to allow you to push the level up to a comfortable level (just don't overload the limiter, as that will probably sound awful and limit the usability of your track...unless that's what you're going for...) but will also allow for unforeseen peaks to happen (as they will).

If you're on a budget, there are preamp/compressor combos that might do the trick.
posted by Jon-A-Thon at 12:00 PM on September 12, 2010


One thing you can do, which is boring but very helpful, is to gain-stage your signal chain. What this means is you work through your signal chain figuring out the correct signal levels at every stage. This prevents you from making the mistake of having to turn up the signal at one stage to compensate for a low signal at a previous stage, or vice versa.

I'm pretty sure there is freeware out there (eg Audigy) which will let you do this stuff with a lot more accuracy than the way you are doing it at the moment.
posted by unSane at 12:35 PM on September 12, 2010


I do believe I better understand getting my level set correctly now.
Waveform looks better. I think you were correct unSane about low levels, and I was trying to make up for that at other stages.
Thank you both for your feedback.
posted by sailormouth at 11:37 AM on September 13, 2010


A nice way to go without using a mixer (which will add noise), is a PreSonus Audio Box USB. It's about $150 and has some great software that comes along with it now. (Also MIDI connections) It has two balanced XLR inputs, and you can plug an unbalanced 1/4 inch input in as well. Then you just add it to the "OPTIONS" device order, and device properties of your Cool Edit. (I have Adobe Audition, which was an upgraded version of what you have... so I'm not sure of where it will show up, but it should automatically appear). Anyway, it has phantom power as well, and it's really easy to set levels. Since it's digital, you want to make sure that you simply just don't go into the "Red". You can always bump up the volume later, or select the entire wave through the "EFFECTS" / AMPLITUDE section to set a more advanced level. I also tend to use the compander section of the compression which is in the same spot EFFECTS/AMPLITUDE/DYNAMICS PROCESSING. Then you can smooth out the wav a bit more... though you should play around with it if it sounds too "pumpy". I've found that this Audiobox by PRESONUS has made it really easy to record. You can also plug your phones directly in, and monitor while you record. You have to set your playback on the multitrack to Audiobox USB. With a little playing around with it, you'll find that it's really easy to multitrack on your computer, with professional results. By the way, AKG has a new "Perception" series condenser Microphone with a one inch diaphram, for a really good price. Comes with a hard case, and shock mount. Perfect for home recording, and sounds good in all situations. I use it for field and studio recording, because I wanted a less expensive mic to take on the road, that I didn't have to worry about replacing if lost, and that sounded good. I was completely impressed by this one!
Hope that helps!
posted by Disbro at 9:16 AM on September 14, 2010


You should be recording as loud as you can without clipping at all stages

I don't at all disagree with unSane on this as a general principle of good practice, but I have lately been pondering a little on the "recording as loud as you can" element. So far as my (limited) technical understanding of engineering goes, it seems that this orthodoxy has its roots in maximising "signal to noise ratio". And that this, in turn, has its basis in drowning out tape hiss (or, I suppose, rumble from acetate laquers before tape came in). In that context getting the signal as hot as possible without clipping is a no-brainer. Particularly if bouncing and overdubbing is a feature of the recording.

But in the digital domain......I'm not so sure this orthodoxy holds. If you're using noisy equipment in the chain pre-tracking, maybe it does. But if you're recording, say, a vocal using equipment with a very low inherent noise floor (like some of the higher-end mics) I can't really see why you'd necessarily need to get the signal up to boiling point. It might even be that cranking things up can, in certain circumstances and perhaps with certain instruments or voices, be a bad idea because it might destroy nuances and subtleties? I have a related bee in my bonnet about what seems to be an increasing tendency for records to be being mixed louder and louder (check out any "remaster" of virtually any album). Why? Just because it's technically possible to mix something to ear-bleeding levels, doesn't mean it should be so.......

These are, as I say, just musings on my part. I know for sure there are others in the MeFiMu commuity that have far more technical knowledge than me and who may be able to elucidate??
posted by MajorDundee at 8:55 AM on September 16, 2010


Yeah - the principles still hold in the digital domain.

There is still an signal-to-noise ratio concern. Any signal system will have an inherent noise floor above which you want to maximize your captured signal, even if it's solid state. Introduce any tube electronics and it shoots right up.

Plus, if you consider that in, for instance, a 16-bit system (for example) you'll have 32768 discrete levels (positive and negative) to represent your signal. There's also a quantization noise that introduced by the process of digitizing the input signal. While it might not be immediately audible, it does actually compound and add up.

Having said that, "recording as loud as you can" is maybe not what I'd personally aim for. I would try and shoot for recording for what sounds good. I agree that louder isn't necessarily better, but there's a difference between reducing the dynamic range (which is the bugbear of over-compression) and recording the highest quality signal you can (which you can always turn down). Sometimes that means placing the mic 10 feet away from you. Sometimes that means cranking the gain (and picking up the room noise) or just playing louder. Sometimes a bit of noise is okay. Depends what you're going for. Take a listen to Arcade Fire's Funeral. To me, that's a lo-fi sounding record. But it's completely eclipsed by the vibe and the performances.
posted by Jon-A-Thon at 10:17 AM on September 16, 2010


When I say 'record as loud as you can' I simply mean that you set your levels so that your peaks are just below zero. This way you are using as much of the representational power of the 16 bits as you can. You're welcome to record softer if you want, but there's no advantage to it (plus all the disadvantages Jon-A-Thon describes.

Moreover, all the digital effects are designed around signals which peak at 0db, so it can be a complete pain if you have a super quiet audio file. You end up putting in gain inserts or normalizing, just so you're not fighting traffic the whole time.

If you record in 24 bit you can be a heck of a lot more relaxed about levels and leave yourself a few dB of headroom.

None of this has anything to do with the loudness wars, which are all about signal crushing via compressors and limiters. Anything final should be mixed to peak just below 0dB, because that's what everything is designed around, and if you don't, you end up forcing people to fiddle with volume controls so they can hear it at the volume they expect.

Also, welcome back, Major!
posted by unSane at 11:04 AM on September 16, 2010


Interesting - thanks to both. It's a kind of order of magnitude thing I think? The more sophisticated the gear gets the more scrupulous the technical examination becomes. What passes for "noise" in a digital recording wouldn't have raised the merest hint of an eyebrow in an analogue environment. And there's a bit of irony in the kind of audio relativity that results in phrases like "lo-fi" to describe recordings - probably all or part-digital - that have a deliberately analogue(ish) flavouring to make them sound "retro". In other words, noise or restricted headroom or some other analogue hallmark has been consciously embedded! If none of that makes sense (and I'm not sure it does entirely) it's because I got to bed at 5.00am after a delayed flight from Cyprus and I'm semi-conscious.

Btw, how's the splendid You & San Fran coming along unSane?
posted by MajorDundee at 12:31 PM on September 16, 2010


I figure you have to set the levels somewhere so you might as well set them to peak under 0dB. But you're right, digital recording is so clean that the noise isn't much of an issue. I've suffered much more frequently from digital clipping than signals being too quiet so I should probably err further on the side of low levels.

I've never really been able to do lo-fi stuff. I'm too much of a nerd and it always bothers me.

Y&SF It's been on hold this week while I rebuilt my studio from the ground up (it's lookin' AWESOME, desperately want to boast about it) and transferred everything over to a shiny new Mac Pro, and tried to get it all working with the new audio interface. But I'm hoping to get all the new stuff tracked next week and rejig the arrangement a bit to make some space for a rather nice new guitar part that arrived in a plain brown package just before you went off on your Greek campaign!
posted by unSane at 1:24 PM on September 16, 2010


I think saying "record as loud as you can without clipping" doesn't mean try and get it as very very hot as possible. - It just means check your meters - because if it clips you are stuffed and if its Rreally really low you can also be stuffed as you are just going to end up adding gain at a later stage and hence truning up your noise / quantization effects.

Sounds like the poster is thinking about it the wrong way. you should be using the Mixer to get everything to the correct level then everything AFTER the mixer i'd set to fixed volumes. - ie the signal coming out of the miixer should be peaking around 0db
posted by mary8nne at 7:59 AM on September 20, 2010


Mary8nne the earlier replys did answer my question, as did yours. I had the mixer output too low everything peaking below 0db, and I was compensating by increasing the line in. I was aware of clipping when I saw the waveform instead of watching the meter like I should have. I was adjusting pretty much everywhere except the mixer output, resulting in some recordings needing to be adjusted in the software and ultimately the final volume changing from song to song. The subsequent discussion in the replys does not bother me. I got the help I was looking for, and I am pleased with the consistent results I am getting.
Again thanks everyone.
posted by sailormouth at 11:53 AM on September 20, 2010


« Older A couple of recording questions re drums/vox   |   Any Chicago singers interested in Siberian folk... Newer »

You are not logged in, either login or create an account to post comments