How to make an album listenable at one volume.
May 5, 2009 6:53 AM
How do I get all my songs to have a uniform volume?
I'm trying to put together something like an album, and an issue I keep having with GarageBand is songs coming out with entirely different volumes. So you'll turn it way up for ones songs, and then your ears will start to bleed as the next one begins.
I'm not sure what causes the variation. I do dick around with the settings without thinking too much, so I'm never sure what the settings were before I changed them. I don't really want to go back and rerecord everything, so is there some way to make everything the same volume in GarageBand? Or maybe another program I can download and load my songs into and the change the volume there?
Thank you!
posted by Corduroy (18 comments total)
6 users marked this as a favorite
I'm trying to put together something like an album, and an issue I keep having with GarageBand is songs coming out with entirely different volumes. So you'll turn it way up for ones songs, and then your ears will start to bleed as the next one begins.
I'm not sure what causes the variation. I do dick around with the settings without thinking too much, so I'm never sure what the settings were before I changed them. I don't really want to go back and rerecord everything, so is there some way to make everything the same volume in GarageBand? Or maybe another program I can download and load my songs into and the change the volume there?
Thank you!
The short answer is "stick a look-ahead limiter on the mix and crank the gain until you hit the desired RMS." The long answer is that it's hard in Garageband (at least the last version I used, which I know is somewhat more primitive than the latest).
Quick and dirty: you'll want to make clean mixes with plenty of headroom. That means when you start mixing, pull all the faders down to -9dB (maybe lower). Nothing goes higher than this (unless you tracked really quietly). Print all your mixes and pull them into something where you can stick a look-ahead brickwall limiter on them. The Apple AU limiter might do this these days, I don't really know. Audacity might have one, too.
Get yourself a copy of RMS Buddy and stick it on your master channel. You want every track to come in around the same average RMS. Once you do that, turn on your limiter for the whole mix, and crank the gain until you hit, say, somewhere between -14dBFS and -11dbFS. This will give you a pretty loud, but not silly-loud, master. You can go louder or quieter, of course, if it sounds better to your ears. But the thing here is that every track should have a very similar average RMS.
There's a lot more to be said about even rough mastering, but that'll at least give you the consistency you're after.
posted by uncleozzy at 9:27 AM on May 5, 2009
Quick and dirty: you'll want to make clean mixes with plenty of headroom. That means when you start mixing, pull all the faders down to -9dB (maybe lower). Nothing goes higher than this (unless you tracked really quietly). Print all your mixes and pull them into something where you can stick a look-ahead brickwall limiter on them. The Apple AU limiter might do this these days, I don't really know. Audacity might have one, too.
Get yourself a copy of RMS Buddy and stick it on your master channel. You want every track to come in around the same average RMS. Once you do that, turn on your limiter for the whole mix, and crank the gain until you hit, say, somewhere between -14dBFS and -11dbFS. This will give you a pretty loud, but not silly-loud, master. You can go louder or quieter, of course, if it sounds better to your ears. But the thing here is that every track should have a very similar average RMS.
There's a lot more to be said about even rough mastering, but that'll at least give you the consistency you're after.
posted by uncleozzy at 9:27 AM on May 5, 2009
This is a closing-the-barn-door solution, but if all your tracks are already done (or you're making an MP3 mixtape for someone with wildly different track volumes), you can use MP3Gain to make all the mp3s the same volume.
posted by Jairus at 7:54 AM on May 6, 2009
posted by Jairus at 7:54 AM on May 6, 2009
1) Good mixes
2) Good mastering
Neither of these are cheap or easy, unfortunately.
posted by chimaera at 8:24 PM on May 6, 2009
2) Good mastering
Neither of these are cheap or easy, unfortunately.
posted by chimaera at 8:24 PM on May 6, 2009
Fascinating question. Easy answer: Songs must be mixed based on frequency content. Broad arrangements should bump higher on the meter than solo acoustic numbers.
I'm a recording engineer so I can't write that without the scientific explanation:
Uncleozzy was right on the money with the RMS suggestion. RMS Meters are slower than digital peak meters and more closely approximate a human's hearing response. There is one major caveat that can be aided with a bit of knowledge in psychoacoustics and acoustic transduction. It most greatly effects groups of tracks that span a wide range of arrangement style from sparse to broad, which seems to fit the tracks of yours that I've heard.
A phenomenon codified as equal-loudness contours should be considered during the mastering phase. Basically, these curves illustrate how the human ear reacts more readily to frequencies found in the human voice. Evolutionarily this makes perfect sense but can confuse the artificial metering tools we use to define loudness of recorded audio. They also explain why music sounds better when you turn it up... or at least when the tracks haven't been compressed to death.
Tracks with a lot of midrange (more or less between 300Hz and 8kHz) should have lower overall RMS level than full range tracks to counteract this effect. A simple example would be a solo voice and guitar track and a dance hall track. If you were to pump the acoustic track up to the point where RMS and peak meters match the acoustic track would sound piercing and overwhelming in comparison to the dance track.
There is also the matter of electrical efficiency in the playback medium. Bass frequencies require more electrical energy than higher ones and are not efficient to transduce into acoustical energy. As a result, a song with heavy bass content will play back much quieter or distorted than one with more mid-high frequency content mastered at the same RMS level. you can also risk overpowering systems which could lead to poor performance and possibly electrical or mechanical failure.
With these things in mind. you should experiment listening on different systems at different volumes. Mastering can be a bit of a black art and I admit I've had hits and misses myself. In the end you may want to send some mixes out to a pro.
posted by dagosto at 2:19 AM on May 7, 2009 [2 favorites]
I'm a recording engineer so I can't write that without the scientific explanation:
Uncleozzy was right on the money with the RMS suggestion. RMS Meters are slower than digital peak meters and more closely approximate a human's hearing response. There is one major caveat that can be aided with a bit of knowledge in psychoacoustics and acoustic transduction. It most greatly effects groups of tracks that span a wide range of arrangement style from sparse to broad, which seems to fit the tracks of yours that I've heard.
A phenomenon codified as equal-loudness contours should be considered during the mastering phase. Basically, these curves illustrate how the human ear reacts more readily to frequencies found in the human voice. Evolutionarily this makes perfect sense but can confuse the artificial metering tools we use to define loudness of recorded audio. They also explain why music sounds better when you turn it up... or at least when the tracks haven't been compressed to death.
Tracks with a lot of midrange (more or less between 300Hz and 8kHz) should have lower overall RMS level than full range tracks to counteract this effect. A simple example would be a solo voice and guitar track and a dance hall track. If you were to pump the acoustic track up to the point where RMS and peak meters match the acoustic track would sound piercing and overwhelming in comparison to the dance track.
There is also the matter of electrical efficiency in the playback medium. Bass frequencies require more electrical energy than higher ones and are not efficient to transduce into acoustical energy. As a result, a song with heavy bass content will play back much quieter or distorted than one with more mid-high frequency content mastered at the same RMS level. you can also risk overpowering systems which could lead to poor performance and possibly electrical or mechanical failure.
With these things in mind. you should experiment listening on different systems at different volumes. Mastering can be a bit of a black art and I admit I've had hits and misses myself. In the end you may want to send some mixes out to a pro.
posted by dagosto at 2:19 AM on May 7, 2009 [2 favorites]
Dagosto - a question for you in your role as an engineer. How come I can't seem to get my stereo masters to sound as loud as modern commercial releases? I always try to get them just shy of clipping (i.e. only going into the red now and again), but even if I really push it they seem to be noticeably quieter than cd's I buy in the shops. It's almost as if the commercial stuff has access to an extra 10dB that us mere mortals are denied......
posted by MajorDundee at 1:29 PM on May 10, 2009
posted by MajorDundee at 1:29 PM on May 10, 2009
I'm not sure what causes the variation.
You are human, not a machine.
So if you're into recording a song, and you really like the way the drums sound verging on distortion, and you build the song around that, it will be different from the next day, when you are singing a ballad quiet and sexy with an acoustic guitar.
Unless, of course, all your songs are the same kind of arrangement, and you still have these big volume changes. Then you have to pay more attention.
Point is, loudness really does vary, a lot, depending of factors which are not immediately obvious. If you want to be a better engineer or producer, you had better try to figure out the controls. Be consistant. Use self discipline. Write stuff down if you can't remember.
If all you want to do is make your music your own way, and all this thinking about technical stuff gets in the way, you had better find an engineer or producer to help you.
posted by Liv Pooleside at 8:37 PM on May 10, 2009
You are human, not a machine.
So if you're into recording a song, and you really like the way the drums sound verging on distortion, and you build the song around that, it will be different from the next day, when you are singing a ballad quiet and sexy with an acoustic guitar.
Unless, of course, all your songs are the same kind of arrangement, and you still have these big volume changes. Then you have to pay more attention.
Point is, loudness really does vary, a lot, depending of factors which are not immediately obvious. If you want to be a better engineer or producer, you had better try to figure out the controls. Be consistant. Use self discipline. Write stuff down if you can't remember.
If all you want to do is make your music your own way, and all this thinking about technical stuff gets in the way, you had better find an engineer or producer to help you.
posted by Liv Pooleside at 8:37 PM on May 10, 2009
It's almost as if the commercial stuff has access to an extra 10dB that us mere mortals are denied......
that's not too far off the mark, actually.
posted by grubi at 7:52 AM on May 11, 2009
that's not too far off the mark, actually.
posted by grubi at 7:52 AM on May 11, 2009
I always try to get them just shy of clipping (i.e. only going into the red now and again)
Are you talking about peak levels, or RMS, though? If you want a really loud master, you're going to have to master through a brickwall limiter, at which point your peak levels become just about irrelevant.
posted by uncleozzy at 8:17 AM on May 11, 2009
Are you talking about peak levels, or RMS, though? If you want a really loud master, you're going to have to master through a brickwall limiter, at which point your peak levels become just about irrelevant.
posted by uncleozzy at 8:17 AM on May 11, 2009
A nice way to achieve adequate results and get the mastering done at the same time is use Izotope's Ozone. It's a farily cheap mastering suite last time I checked, the guide that they offer will teach you quite a bit about mastering even if you don't use that suite.
posted by bigmusic at 11:52 AM on May 11, 2009
posted by bigmusic at 11:52 AM on May 11, 2009
uncleozzy - many thanks for that. You've exposed my abject and miserable ignorance of all things technical - I actually have no idea what I'm talking about...peak levels...RMS.....wha?? And I haven't the foggiest idea what a "brickwall limiter" is. Obviously, I'm not an engineer or, frankly, particularly interested in that side of things. I only do it because I can't afford to pay someone else to sit around for hours while I agonise over whether a Strat would be better than a Les Paul on that entirely superfluous 2-second fill at the end of the bridge. I tend to just go with what sounds good to me. I do take notes when I get lucky and something works - but my general approach is very much hit and miss so far as engineering goes. So what is a "brickwall limiter", how do I recognise one in its natural environment, and is it edible??
posted by MajorDundee at 2:26 PM on May 13, 2009
posted by MajorDundee at 2:26 PM on May 13, 2009
I am anything but an expert, but I'll try to give some explanation, MajorDundee.
Let's start at the (sort-of) beginning. Let's say you've got a mix of a song that you're happy with. Only problem is, it's not very loud. Let's say it peaks somewhere around -6dB. You could just turn it up by 6dB, have it peak at or near 0dB, and be done with it (it sounds like this might be what you're doing). But it still wouldn't be very loud.
Even though you're peaking at 0dB, unless you've got a ton of compression on your mix, the bulk of your audio is actually much softer than that (maybe the peaks are mostly kick or snare hits, for example). Even with the right amount of compression, your mix is likely to sound very quiet. On the one hand, this is a good thing: the wide dynamic range gives you greater detail in your mix. On the other hand, though, in the real world, people listen to music in cars and on trains and so forth, and nobody can hear the quiet bits. Enter the limiter.
A limiter is essentially a high-ratio compressor. Think of a compressor like your mom with her hand on the volume knob: when something gets too loud (the threshold), she turns it down by a specified amount (the ratio) after a certain amount of time (the attack), and eventually returns it to its initial position (the release). So the loud parts are a little, or a lot, less loud. A limiter has a very fast attack and very high ratio.
So you've got this mix where the RMS--that is, the average, perceived loudness--is relatively quiet, but you can't make it any louder without inducing clipping (going "above" 0dB). So you take this brickwall limiter and stick it on the mix. As noted above, get yourself an RMS meter and look at it (RMS Buddy is great, since it will show you RMS over time). Let's say your RMS for the entire track is -18dBFS. That's fairly quiet. What you really want is something more in the -14dBFS to -11dBFS range. So you use your limiter to keep the peaks below 0dB (I usually limit to -0.3dB) and turn up the gain by 4-7dB. You've really got to use your ears here. As dagosto notes, what's going to happen here depends a lot on the makeup of your track, and there are no hard-and-fast rules.
As far as identifying a brickwall limiter, the standard limiter in most DAWs will have a brickwall mode (sometimes called "no overs"). Logic will hard-limit in its standard compressor, in the limiter, and in the adaptive limiter (which has a longer lookahead for somewhat-cleaner performance). Software limiters are powerful tools, and you've got to be careful with them, or you'll wind up with a loud, distorted mess. But, carefully applied to a good mix, they're invaluable for squeezing out a few extra dB.
posted by uncleozzy at 7:37 AM on May 14, 2009
Let's start at the (sort-of) beginning. Let's say you've got a mix of a song that you're happy with. Only problem is, it's not very loud. Let's say it peaks somewhere around -6dB. You could just turn it up by 6dB, have it peak at or near 0dB, and be done with it (it sounds like this might be what you're doing). But it still wouldn't be very loud.
Even though you're peaking at 0dB, unless you've got a ton of compression on your mix, the bulk of your audio is actually much softer than that (maybe the peaks are mostly kick or snare hits, for example). Even with the right amount of compression, your mix is likely to sound very quiet. On the one hand, this is a good thing: the wide dynamic range gives you greater detail in your mix. On the other hand, though, in the real world, people listen to music in cars and on trains and so forth, and nobody can hear the quiet bits. Enter the limiter.
A limiter is essentially a high-ratio compressor. Think of a compressor like your mom with her hand on the volume knob: when something gets too loud (the threshold), she turns it down by a specified amount (the ratio) after a certain amount of time (the attack), and eventually returns it to its initial position (the release). So the loud parts are a little, or a lot, less loud. A limiter has a very fast attack and very high ratio.
So you've got this mix where the RMS--that is, the average, perceived loudness--is relatively quiet, but you can't make it any louder without inducing clipping (going "above" 0dB). So you take this brickwall limiter and stick it on the mix. As noted above, get yourself an RMS meter and look at it (RMS Buddy is great, since it will show you RMS over time). Let's say your RMS for the entire track is -18dBFS. That's fairly quiet. What you really want is something more in the -14dBFS to -11dBFS range. So you use your limiter to keep the peaks below 0dB (I usually limit to -0.3dB) and turn up the gain by 4-7dB. You've really got to use your ears here. As dagosto notes, what's going to happen here depends a lot on the makeup of your track, and there are no hard-and-fast rules.
As far as identifying a brickwall limiter, the standard limiter in most DAWs will have a brickwall mode (sometimes called "no overs"). Logic will hard-limit in its standard compressor, in the limiter, and in the adaptive limiter (which has a longer lookahead for somewhat-cleaner performance). Software limiters are powerful tools, and you've got to be careful with them, or you'll wind up with a loud, distorted mess. But, carefully applied to a good mix, they're invaluable for squeezing out a few extra dB.
posted by uncleozzy at 7:37 AM on May 14, 2009
MajorDunde - Sorry I haven't checked back on the thread in a bit.
uncleozzy is right on about brick-wall limiters.
I would also add, to expand a bit on my previous post, that there are tools that control levels of different frequency bands (low, low mid, mid, high) used to further increase loudness. A multi-band limiter compresses each band separately. To use uncleozzy's example, it is like several moms controlling the volume on different parts of the music. It is usually used to even out the sound across the entire range of frequencies.
In a dense mix a general rule is that lower frequencies need more compression than higher ones to create more loudness. Plus high frequencies can sound washy when really compressed and usually sound better with faster attack and release settings. You can get really clean sounding loud tracks if you use multiband compression like this.
I use it sometimes on a mix buss or even every once in a while on an individual track. Most pro records you hear will have some form of multiband on them. It can be a little daunting to use since it is like five or six compressors and an eq all at once. You may need to experiment with it.
posted by dagosto at 2:30 PM on May 15, 2009
uncleozzy is right on about brick-wall limiters.
I would also add, to expand a bit on my previous post, that there are tools that control levels of different frequency bands (low, low mid, mid, high) used to further increase loudness. A multi-band limiter compresses each band separately. To use uncleozzy's example, it is like several moms controlling the volume on different parts of the music. It is usually used to even out the sound across the entire range of frequencies.
In a dense mix a general rule is that lower frequencies need more compression than higher ones to create more loudness. Plus high frequencies can sound washy when really compressed and usually sound better with faster attack and release settings. You can get really clean sounding loud tracks if you use multiband compression like this.
I use it sometimes on a mix buss or even every once in a while on an individual track. Most pro records you hear will have some form of multiband on them. It can be a little daunting to use since it is like five or six compressors and an eq all at once. You may need to experiment with it.
posted by dagosto at 2:30 PM on May 15, 2009
uncleozzy and dagosto - very many thanks, I appreciate the advice. I'll have to check out what I can get that will work with my DAW (Yamaha AW2400). Main criterion for me will be simplicity of use - I'm not a "techie" and if it's too much of a pain in the arse to use it'll stay in its box. Thing is, by the time I've written and recorded a track I'm beginning to get pretty bored with it and want to move on to something else, so if I have to then spend hours farting about with software and magic boxes....well...it ain't gonna happen. It's a pity I can't e-mail my mixes (or multi-tracks) to guys like you - who know what they're doing - to see what results. The WAV files are way to big for that. Thanks again.
posted by MajorDundee at 4:07 PM on May 15, 2009
posted by MajorDundee at 4:07 PM on May 15, 2009
I would also like to add here that using a brickwall limiter or any other heavy dynamic processing on your final mix should only be done if you are not going to have your project professionally mastered.
If you are going to send it out to be mastered by a professional, let them do it. They have much better ears and equipment than you do and can make it sound much better. If you already squeeze out the dynamic range, they will not have much to work with.
posted by chillmost at 5:01 AM on May 20, 2009
If you are going to send it out to be mastered by a professional, let them do it. They have much better ears and equipment than you do and can make it sound much better. If you already squeeze out the dynamic range, they will not have much to work with.
posted by chillmost at 5:01 AM on May 20, 2009
Better equipment - no doubt, better ears - hhhmmm, not so sure about that. I've had stuff "professionally" mastered and been a member of two bands with record deals. I've seen good mixes f*cked up at the mastering stage more than once. But generally you're right re giving them stuff that's not had the life compressed or EQ'd out of it. Just make sure you're there when they do it....
posted by MajorDundee at 12:57 PM on May 20, 2009
posted by MajorDundee at 12:57 PM on May 20, 2009
As an amateur mastering engineer let me say that the fuck ups usually come when the mix is terrible. Mixing is often approached as a subjective practice. This is good to some extent but physics trumps all. A misunderstanding of the physics makes for bad mastering.
The biggest problem today is that it is not considered proper to reject a mix. A mastering engineer should be comfortable sending a mix back with a list of critiques. Good mixes is the only way to get good masters.
posted by dagosto at 12:31 AM on May 22, 2009
The biggest problem today is that it is not considered proper to reject a mix. A mastering engineer should be comfortable sending a mix back with a list of critiques. Good mixes is the only way to get good masters.
posted by dagosto at 12:31 AM on May 22, 2009
Quite right dagosto - to use an apt English phrase "You can't polish a turd"! What I was getting at was if you do have a good mix - i.e. something that sounds good and is nicely balanced etc - then a cloth-eared mastering engineer can make it sound too bassy or trebly or whatever. Not sure about your "physics trumps all" assertion - I guess it's the age-old science v art debate. I'm going to start a new string on this: "To what extent is recording a science or an art? - discuss".* Should be interesting.....
*For the purposes of discussion assume that "recording" includes the writing, arranging and performing of a piece of work
posted by MajorDundee at 10:47 AM on May 22, 2009
*For the purposes of discussion assume that "recording" includes the writing, arranging and performing of a piece of work
posted by MajorDundee at 10:47 AM on May 22, 2009
You are not logged in, either login or create an account to post comments
2.) Create a new song.
3.) Import all songs as separate tracks. Set the master track compressors and volume. This is where you can set your absolute limits.
4.) Pick two tracks and adjust them to be the same volume, using the solo functions and individual track eqs and compressors.
5.) Use one of the tracks you just adjusted and then select a different track and match them for volume.
6.) Keep repeating 4 & 5 until you're done.
7.) Do it all over again, selecting different combinations.
8.) When completely satisfied, export each track individually as a "master". You can also do kooky segue ways between tracks at this point.
9.) ...
10.) Profit!
posted by dobie at 8:30 AM on May 5, 2009