Why are my songs so soft!

Jayhell

Why are my songs always so soft? After I record a song I usually do a bit of mastering (not that I know enough about the subject to actually call it mastering) and then I convert it to mp3. But compared to other songs they are always so soft, how do I get them nice and loud without compressing the life out of them?

greetings

Meters.

Firstly make sure you are recording at 24 bit, then you don't have to record so hot as to get away from the noise floor. Aim to get your recording input levels to about -12dB (in the digital domain). You'll want the loudest peak of your mix to be from about -5dB to -3dB, and then in mastering you aim to get it to a max of -0.01dB.

NEVER let it clip. This causes you to lose bits of data. Just watch your meters.

Mo-Facta

Metering is not really the issue.

This is a problem that many post about on the internet and the fact of the matter is that a loud master is a result of many aspects that starts at the recording phase. The first issue is that loudness should not the be goal. A great recording should be. A good arrangement, a strong mono phantom image, a good frequency balance and many other factors contribute equally as well. Also, controlling the dynamic range in the recording and mixing phase, especially with pop/rock music, can go a long way.

These days I'm easily reaching an average RMS of about -10 to -8dBfs without taking more than 1 to 3 dB off the highest peaks at the limiting phase. It's just something you learn over time and comes naturally when your records become better.

As always, there's just no magic bullet for anything in this craft.

Cheers ?

Arno-West

What do you use for mastering? If Wavelab, do the following:
1. Drag your mixed (at no more than -3db) track into Wavelab.
2. Double click on the Wav display.
3. Normalize it to -0.1db.
4. Next render a Mastering Compressor onto whole track (I use the Waves RComp with the Opto preset for this).
5. Then, under the Process tab there is a Loudness Normalize option. Choose the "Catch loudness from track" option and render the track 2db louder than what the software suggests.
6. Finally do a Global Analysis and fix all audio errors as analyzed by Wavelab.
Your track should be nice and ballsy now. Remember to save the track and when converting to Mp3 never go below 192kbps.
All of the above works for me like a charm. Others might do things differently.
If you dont use Wavelab, sorry cant help...

Arno-West

PS. When mixing, set the kick drum and bass as main instruments then bring in everything els just underneath those two. The exception is the snare drum which should (99% of the time) be on the same level as the lead vocal which in turn should sit just above the kick and bass. Those 4 tracks give your track a foundation, the rest add the "meat". Oh, and dont be scared of the mids on guitars. Rather add mids than volume to get them to sit in the mix.

AlanRatcliffe

I used to try avoid compression as much as possible, but that's not really practical. I find I get better results by normalising every track and then I add lots of instances of fairly subtle limiting or compression (mostly just controlling the peaks). Some to individual tracks, some to submixes and a few db of multiband compression to the stereo mix. That way I can maximise the volume while retaining fairly natural dynamics.

Mo-Facta

Arno West wrote: What do you use for mastering? If Wavelab, do the following:
1. Drag your mixed (at no more than -3db) track into Wavelab.
2. Double click on the Wav display.
3. Normalize it to -0.1db.
4. Next render a Mastering Compressor onto whole track (I use the Waves RComp with the Opto preset for this).
5. Then, under the Process tab there is a Loudness Normalize option. Choose the "Catch loudness from track" option and render the track 2db louder than what the software suggests.
6. Finally do a Global Analysis and fix all audio errors as analyzed by Wavelab.
Your track should be nice and ballsy now. Remember to save the track and when converting to Mp3 never go below 192kbps.
All of the above works for me like a charm. Others might do things differently.
If you dont use Wavelab, sorry cant help...

I wouldn't do this, but hey, it might work for him. Sorry Arno, but what you have suggested is bad practice. How can you give this advice without possibly knowing what the source track sounds like?

The Wavelab Loudness Normalizer merely uses the Peak Master plugin that comes standard in Wavelab to process the waveform and therefore uses brickwall limiting with a combination of a soft clip function. Truth be told, it's not the BEST sounding loundess maximizer although I know of a few pros that like it. I personally don't. The method you have suggested is a hit or miss thing as his track may or may not stand up to this type of extreme processing.

Cheers ?

Jayhell

Cool, I dont use Wavelab, I use Cubase for mastering as well. I'll try all the above suggestions and see what works. I've never used normalization, I'll check that out.

AlanRatcliffe

Normalising is not essential - especially if you are capturing the right levels when tracking.

I normalise because it makes sure all my tracks are in the same level range for any compression in the track inserts to operate predictably with the presets I've saved over time. I still go in and tweak settings later, but it lets me quickly insert compressors with some decent default settings.

Arno-West

Mo Facta wrote:
Arno West wrote: What do you use for mastering? If Wavelab, do the following:
1. Drag your mixed (at no more than -3db) track into Wavelab.
2. Double click on the Wav display.
3. Normalize it to -0.1db.
4. Next render a Mastering Compressor onto whole track (I use the Waves RComp with the Opto preset for this).
5. Then, under the Process tab there is a Loudness Normalize option. Choose the "Catch loudness from track" option and render the track 2db louder than what the software suggests.
6. Finally do a Global Analysis and fix all audio errors as analyzed by Wavelab.
Your track should be nice and ballsy now. Remember to save the track and when converting to Mp3 never go below 192kbps.
All of the above works for me like a charm. Others might do things differently.
If you dont use Wavelab, sorry cant help...
I wouldn't do this, but hey, it might work for him. Sorry Arno, but what you have suggested is bad practice. How can you give this advice without possibly knowing what the source track sounds like?

The Wavelab Loudness Normalizer merely uses the Peak Master plugin that comes standard in Wavelab to process the waveform and therefore uses brickwall limiting with a combination of a soft clip function. Truth be told, it's not the BEST sounding loundess maximizer although I know of a few pros that like it. I personally don't. The method you have suggested is a hit or miss thing as his track may or may not stand up to this type of extreme processing.

Cheers ?

Well, he wanted to know how to get his songs louder. We all know every song is different and we all have certain ways (quite a few) of doing things. I gave him a quick way of making the songs louder. If you lose some dynamics along the way, so be it. He said he converts to Mp3 so he's gonna lose some quality anyway.

greetings

Some very good points. The way I took it was that he was not mixing to a loud enough reference level, or maybe didn't understand that. But I see that that was not the case.

Mo-Facta

Alan Ratcliffe wrote: Normalising is not essential - especially if you are capturing the right levels when tracking.

I normalise because it makes sure all my tracks are in the same level range for any compression in the track inserts to operate predictably with the presets I've saved over time. I still go in and tweak settings later, but it lets me quickly insert compressors with some decent default settings.

Sure, normalizing is a tool just like any other. Knowing when and why to use it is, as always, key.

Wavelab actually has a cool feature in Audio Montage mode called "Meta-Normalizing" that let's you normalize any selected tracks to a number of different configurable parameters. For instance, it will let you loudness normalize each track under a certain peak level in preparation for processing. So, for instance you can specify the amount of headroom you require for processing while also making sure that the relative loudness of each track is consistent. Of course, it's not perfect and often requires that you listen for gaps in the analyzing process because it's still quite difficult for computers to accurately determine how the ear perceives loudness in the real world.

Cheers ?

BMU

My answer to this turned out to be the same as many other things: throw a bit of money at some genius software geeks.

Specifically, oZone iZotope (Or iZotope oZone... I can never remember which is the company and which the product. Which just shows how deeply I care about the finer details haha.)

Insta-loudness (like, by a LOT) PLUS makes everything sound insta-awesome with two little sliders. Woohoo.

Of course oZone/izotope is also a serious tool if you want it. I do also dabble in EQ matching, another lazy man's shortcut to sounding somewhat like a Sneap production.

I understand and respect the whole each-situation-is-unique-and-very-complicated drill, and that's 100% applicable for studio peeps, or anyone interested in it. But for guitarists/songwriters who want to generate demos that sound pretty much good enough, products like this are a heavensend.