Hello -
Here comes another “how do you do this” question… When I get my tracks / songs about ready to put on a cd, I notice a slight variation from song to song in terms of one is a bit louder than another. I tried the normalization route and that didn’t seem to get songs in sync with each other like the professionally done CDs sound.
Soooo, how do you folks get your CDs at the same volume level?
Thanks!
When I mix down, I use a mastering limiter on the master channel.
One of the easiest to use (and free, too!) is the Kjaerhusaudio Classic Series Mastering Limiter.
In order to use this technique to its full potential, leave the mastering limiter out of the mix until the mix sounds exactly as you want it to sound! This is very important!! Small differences in levels, EQ settings etc. may be very hard to determine when the mastering limiter is strapped over the master channel. Leave it out until the mix is ready, and then put it in using the default setting. This should keep all levels even between songs.
Regards, Nils
I use the n-Track multiband compressor for most all my mixes.
The 2.4 preset is close for almost all I do.
Try that and see how it sounds for you by turning it on and off during play - tweak it as needed.
I will pass on some advise I got:
Unfortunately, there is no software that will bring all audio to the same “Perceived” level. Looking at the wav form or maxing the levels will not make every song sound like it is the same volume - not unless they are all of the same type (read: loud, heavy, with no dynamics). A good collection of songs has a level that is comfortable to listen to, you don’t want people to have to turn it up or down for each song, but each song may actually have a slightly different maximum volume. The human ear hears different sound differently - is there a lot of presence, is it an “open” sound one just aguitar and singer, or is the singer down in the mix of drums and electric guitars, and so forth. It’s an art. Get the levels about even on the meters, then listen to the playbacks and adjust accordingly.
Thanks! I appreciate all your ideas. I’ll letcha know how it turns out!
r
In addition to all of the above, as I render a mixdown of each song from n-Track, I look at it in Goldwave and check it with the “Match Volume” feature of Goldwave. Match Volume measures the perceived loudness of a song based on the average level, not the peak level like normalizing does. I target between -16.0 and 16.5dB. If the song falls out of this range, I adjust the mastering limiter in n-track accordingly and re-render the song. My resulting CDs have tracks that are all the same volume.
Quote: (bax3 @ Apr. 30 2008, 2:32 PM)
I will pass on some advise I got:
Unfortunately, there is no software that will bring all audio to the same "Perceived" level. Looking at the wav form or maxing the levels will not make every song sound like it is the same volume - not unless they are all of the same type (read: loud, heavy, with no dynamics). A good collection of songs has a level that is comfortable to listen to, you don't want people to have to turn it up or down for each song, but each song may actually have a slightly different maximum volume. The human ear hears different sound differently - is there a lot of presence, is it an "open" sound one just aguitar and singer, or is the singer down in the mix of drums and electric guitars, and so forth. It's an art. Get the levels about even on the meters, then listen to the playbacks and adjust accordingly.
That's a good point there.
Actually when I've had stuff professionally mastered, it's been more important to get the songs to sound right together and to get their transitions sounding right rather than to get everything at the "same" level. "Use your ears" I think is the best advice.
I agree with Bax and Mark, and Vivona has a good point as well.
There is no magic bullet; though there are tools like the one in Goldwave to do 80% of the job.
Note that Vivona says he “looks” at it in GoldWave – he doesn’t necessarily let the feature try to do the whole job.
Getting the volume levels right is a part of what we call “mastering”.
Mastering is a big subject; like mixing it’s an art that takes a good understanding of technical fundamentals.
Normalization is definitely the wrong tool, because, as mentioned above, it only looks at peak levels.
Peak levels don’t have much to do with perceived loudness.
The next best approximation is RMS levels, which can be found in various metering plugs, and sometimes built into DAWs.
(I’d have liked to have seen it built into n-Track – it may be with later versions than I’ve uised.)
RMS levels tell you the energy level os it’s a much better approximation than peak levels, and I definitely use RMS levels as a guideline when mastering a collection.
However, RMS doesn’t take any psychoacoustics into account.
The next step up is VU metering, but unfortunately, there are lots of plugs that call themselves VU meters that aren’t, so caveat emptor.
VU meters are another relatively simple thing mechanically – it was the best they could do with simple analog parts and an electromechanical needle to make a meter that’s basically a flawed RMS meter, where the flaws make it closer to our ears than a perfect RMS meter would be.
No doubt there are plugins available with highly refined psychoacoustic adjustments built into them, though none that I’m aware of (i.e., cheap or free).
If anyone knows a good one, please post!
Note that I’m only talking about meters here.
The idea being that you choose a target level for the average in your CD, but each track may be higher or lower, depending on the dynamics of the CD.
A quiet song probably should have a lower RMS or VU level than a super blaster.
Here’s what I do, using n-Track.
First, I mix each song.
Next, I make sure that most of each song peaks near zero.
If there are a few peaks that go over zero, it’s OK if you follow my method.
People will tell you NEVER to peak over zero.
It’s good advice, for a number of reasons, but in this case it really won’t matter if a few peaks go over zero.
If lots of peaks go over zero, then you can’t possibly be hearing what you’re mixing, so turn the master gain down!
I mix the song to 32-bits.
(You can use 64 bits if you prefer, and don’t mind wasting disk space.
I think N supports 64-bit mixdown these days.)
Then I load each song into a single n-Track project (“song file”).
I leave them on their respective tracks, but slide them right, into the order I want them on CD.
I load my favorite compressor/limiter plugin on the master channel.
Spot checking in all of the songs, I set the parameters so the whole collection sounds reasonable, without being too fussy about it.
I also load any ambience plugins that I want applied to the whole CD here, like master reverb and/or tape saturation.
I also load an RMS metering plugin on the master channel, making sure it’s last in the chain.
I pick a nominal RMS level that I think is appropriate for the genre.
For my music, I like somewhere from -12 to -9 dBFS.
For wall-o-sound rock / heavy metal, I might go as high as -6dB.
Modern mixes often squeeze it up to -3, but I highly recommend against that!
A, it sounds like sht, and B, they’re experts and doing a REALLY good job and you’re not likely to do nearly as well so it’ll sound like even worse shit.
Frankly, even -6dB is very aggressive.
For beginners, I recommend no higher than -12dB, but keep in mind that at this level, when people put your tunes in a mix with commercial stuff, it’ll sound quieter (for the louder genres).
I’m assuming the first cut on the CD is a reasonably “up” one – usually a good choice.
Not the biggest screamer, but one you’d want louder than average.
(For one thing, you’re doing your listeners a favor.
Every time someone has to get up to turn your CD down, you should lose five cents!)
I adjust compressor and track fader to get the target level on most of the song.
Then I do whatever’s necessary to tame any roque peaks, using compression and/or limiting – on the song’s track if it’s likely to be an unusual situation, or on the master channel if this song is typical.
Note that with a limiter in the master channel, I might not see peaks that are being squashed, so I often turn off the limiter to see what the damage is.
When pushing really high nominal levels like -6dB, I’d do this with everything at -9 dB because at -6 the limiter will be busy no matter what.
All this time, I’m using my most important tools: my ears and my brain.
Avoid trying to fly on instruments.
This is why good monitors are very important.
I don’t have good monitors, but do as I say, not what I do!
When I’m happy with the level and the sound of the first song, I do as Mark says and start paying more attention to the transitions; so each song sounds right following its predecessor.
I note the RMS levels I get, mostly to make sure I don’t get caught in an upwards or downwards spiral.
That is, I make sure that most of the CD is at my nominal target.
Quiter tunes are lower, blasters are louder (but not much!)
When I’m done working all the transitions and my CD is mostly at my target level, I then listen to each song carefully, listening to bad stuff like compressors pumping or too much limiting (causing a squashed sound).
When I’m done, success is relatively easy to judge.
If it’s good and I’m playing it on a good sound system, I’ll always want to turn it up louder.
Even if it’s pretty damn loud.
If I find myself wanting to turn it down, that’s a clue that ear fatigue is setting in, probably because of too much compression, or too much distortion.
If so, back to the starting gate.
Now, NOTE WELL: When mixing and mastering, do most of your work at relatively low sound pressure levels (low volumes).
The louder you mix, the better it sounds.
However, this is
(a) hiding weaknesses – you’ll still hear important things that will be lost at low levels.
(b) fatiguing your ears, whose response changes over time when exposed to high levels.
Working at high levels is a recipe for disaster, not to mention hearing loss.
Also, take periodic quiet breaks.
Get yourself a Radio Shack SPL meter, set it to “C” mode, and make sure your typical mixing level is below 85 dB.
Or google Robert Katz and see what he says about monitoring levels, and do that.
HTH!