[linux-audio-user] Fwd: [Jamin] Re: soft clip: Achieving Gain, inconsequential overloads

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



"Hi,

Maybe some folks on this list have insites into the
following conversation between Steve and myself. I
need all the help I can get. The last three paragraphs
are a topic for which I am woefully ignorant."

Hiya. I do the occasional master for releases, and "make it louder" is an all too common request. Whether it suits the track, the place in the album or the recording does not matter, they just want it LOUD. It's sad, but he who pays the piper....
Yes, you have to mash the audio (as you have found), and it kills me to hear carefully recorded tracks being brutally handled, so the idea is to do it as kindly as possible. :)

Here's some rambling about how I go about it.
First, where is all the energy in the track? What is taking up all your headroom? 

Subsonics can really take up headroom, so going back to the mix and high passing any tracks that don't need bass end can give you more space to work with. Starting at 20hz on most tracks and work up from there.

Then, put a fast limiter across your main outputs, set it so it's fairly hammering the track, and listen for the moments when it really ducks.
Look what's going on in the track at those points and automate/eq those parts. Much of getting a 'loud' track is in the mixing.
Pretty much any limiter will work here. You *want* it to be offensive. :)

My chain for mastering is normally - EQ/Exciter -> Comp -> Multiband Comp-> Limiter -> Dither.

The first eq gets the general shape. Don't move on till you are happy. Try a steep highpass at 10hz or whatever here too. Creep it up till you can hear it then back off.
Getting this right really depends on your monitoring. If your monitors can't do anything below 40hz and you are mastering electronic/dance be careful as there can be a lot going on down there. Listening from outside the control room with the door open seems to help here. I have no idea why.

The next compressor is optional. If the tracks fairly dynamic I'll use one. It will subtle though. It's a good way to make the multiband more predictable when going for volume too.
Perversely, setting a very low threshold works here. You'd think that would mash the track, as the compressor is working nearly all the time, but as the ratio is so low (1.2:1 or whatever) combined with a long release (1 sec or greater) it just smooths out the dynamics before it hits the multiband. Soft knee is essential here. You should not hear this 'working' at all ideally. An opto comp hear will let some peaks through but still bring up the general ambience in a pleasent way.
 
You *want* to be using a multiband comp for volume. 

Try finding the range where the vocals are and fitting your middle multibands around that. You need to be able to solo bands to do this ideally. This means that the bass end interfere won't with the vocals, so they don't duck when there is heavy kick or bass. If the vocals remain up front you are half way there. If you have more bands, try finding the space between the kick and bass. Don't go too heavy on the top end bands, they should still have punch. 

As a rule of thumb, I play the track through the multiband, and play with the thresholds so each band is doing no more than 3-4db reduction at their respective loudest points, perhaps less on the upper bands. 
Then, play with the input level to the multiband to see what you can get away with.

You can get away with more compression in the low end than the high end, and the more you control that low end the less the limiter has to try and control your kick/bass etc.
Also, using very fast attack/release in the upper bands can work well, but don't kill the snare. Try soloing the low band and reducing the attack release until you begin to hear distortion, then back off a long way.

Now the final limiter.
Ideally, the multiband is doing most of the work and the final limiter is not working hard. If you try and get all your level control out of that limiter it's going to be working too hard and everything will mash up. I can not get the same amount of clean limiting out of any apps on Linux as I can out of a Finalizer or Waves L2. That's just Life. :) If you are doing anything more than 6db reduction here, something is wrong earlier in the chain. 

Overloads on your final master.
Take care. If there are too many flattened peaks you can end up with a CD that sounds OK on your player, but starts getting unpleasent on older/crappier ones. I have had this happen, even with stuff that is not actually digitally clipped, just heavily limited. You should not need to do hard digital clipping at all, ever. You will end up with high frequency hash that will make radio station's limiters do odd things and some CD players cry. If you are not getting the volume you need in other ways, buy or rent a finalizer and stick it through that. It's got automatic wizards to get you most of the way there. It's what everyone else does.

If you do want to drive it a bit, try mastering to a decent 2-track tape and pushing that. Depending on the machine it'll get rid of peaks wonderfully.
Just mastering on to tape at a normal level and re-recording+normalising the result can get you a few extra dbs without any obvious change to the sound.

A little clipping on an analog desk is not a problem. If it sounds good, go with it. :)

Just remember to keep an un-mashed master for when the current passion for square-waved CDs dies away. Your clients will thank you for it someday. 
But really, a Finalizer does this kind of thing really well. They don't sound great but they do the job for volume.


[Index of Archives]     [Linux Sound]     [ALSA Users]     [Pulse Audio]     [ALSA Devel]     [Sox Users]     [Linux Media]     [Kernel]     [Photo Sharing]     [Gimp]     [Yosemite News]     [Linux Media]

  Powered by Linux