How to make a song louder without quality lossand clipping?

[Master Mi]

I have a problem with one of my remix projects.

I've reached the maximum of my volume level in my DAW mixer - the volume level is just below 0 (above this level begins the yellow section - not sure what this is for, it's just 3 further dB - above this is the red section that shows clipping in the mastering suite of my DAW).

So I left the master volume in the whole track below 0/below yellow.

But after exporting und uploading the remix on soundcloud and youtube I sensed that the volume was not loud enough after comparing it to other remixes and songs at those sites.

I'm not sure if this problem has to do with the instruments I've chosen in the remix project or with the basis of the whole DAW.

Between the instruments of the highest volume level and the instruments of lowest volume level is a range of about 10 dB - plus the 3 dB in the yellow section would make 13 dB.

But the volume configuration of the single instruments are quite perfect for the track - I really don't want to change these.

Do you know if there's a possibility right within a DAW (or an a special program for post production/volume control) program to make the songs louder without clipping or a loss of sound quality?

[theshaggyfreak]

Do a Google search on the basics of mastering. You'll find several articles that should help you out. It's a process that takes some practice but it's quite doable once you get the ear for it.

I generally never mix louder than -6db. My mixes come out of my DAW somewhere between -12db and -6db. I then master it using Wavelab Elements (you can do this in just about anything, though) using a parametric EQ, a compressor, and a limiter. After that's done, my song will peek out at 0db but I make sure not to squash things so much that the dynamics are lost. I'm mostly using plugins from Melda to do all of that work.

If you want a really in depth read on mastering, Mastering Audio by Bob Katz is the book to get. I learned a lot from it and still use it as a reference.

Edited June 20, 2014 by theshaggyfreak

[zircon]

The abbreviated answer is that you use a limiter plugin. A limiter prevents the audio from going above a certain level (like 0db). Thus, no matter how much you crank up the volume before the limiter, it will never clip. However, "sound quality" is another issue altogether.

All the frequencies within a sound - a drum hit, bass line, electric guitar riff - contribute to its overall volume. Let's take a nice electric guitar sound that you want to use as a lead. Depending on how the sample was recorded, it might have a lot of frequency content in the sub-250hz range. While those frequencies DO contribute to the VOLUME of the guitar, we (as listeners) might not care about them for that particular instrument. So, you use an EQ to remove everything below 250hz for that guitar. Now, the important frequencies of the guitar sound just as loud as before (everything ABOVE 250hz), but the overall volume of the guitar is actually lower. Hooray!

Therefore, by carefully EQing your instruments, you can reduce volume for those frequencies that don't matter, while retaining the important frequency information for any given instrument. In my experience, for MOST instruments other than bass & kick sounds, this means reducing the <250hz frequencies. This will usually have the added benefit of making your song sound "cleaner", better-mixed and more balanced... less "muddy".

Related to this, your choice of instruments and notes matters. Using, say, 5 different basslines in the low range, all at the same time, can clog up your mix. You're using a lot of volume ('headroom') but the song won't sound full at all. On the other hand, by using a variety of instruments across the entire spectrum of notes & frequencies, things will sound way better. A well-mixed, well-arranged song is key to getting more volume.

[Master Mi]

Man, this could be the reason!

I remember that I have configured every instrument layer in the mixer with an EQ preset (Hi-Fi-Sound with raised volume of basses/low frequencies and high frequencies).

I will try to remove these settings and put this EQ preset maybe better over the whole track afterwards.

[timaeus222]

Man, this could be the reason!

I remember that I have configured every instrument layer in the mixer with an EQ preset (Hi-Fi-Sound with raised volume of basses/low frequencies and high frequencies).

I will try to remove these settings and put this EQ preset maybe better over the whole track afterwards.

I would not use EQ presets at all. It's highly unlikely that there are supposedly "universal" presets that actually work for everything, because every sound is distinct. Instead I would use my ears and see what changing EQ at particular frequencies does to the sound and ask for feedback, learning in the process. Using presets does not help your learning if you just use them and do not edit them.

Edited June 20, 2014 by timaeus222

[theshaggyfreak]

I would not use EQ presets at all. It's highly unlikely that there are supposedly "universal" presets that actually work for everything, because every sound is distinct. Instead I would use my ears and see what changing EQ at particular frequencies does to the sound and ask for feedback, learning in the process. Using presets does not help your learning if you just use them and do not edit them.

It really helps to have an EQ with a built in spectrum analyzer. Of course, using your ears is always best but an analyzer sure can help figure out where the annoying frequencies are poking out at. Learning how to go through this process will always be better than any setting.

[Master Mi]

I' ve just checked out the EQ thing.

It lowers the at most possible master volume (right below clipping) a lot - just by ear I would says it makes a difference up to 30 % of loudness (if you put an EQ over every instrument or not).

That's the gain I would need to make it sound like the other songs.

@timaeus222: So you would everything try to do to avoid equalizers (or just the EQ presets?) in general and make the whole sound just by the instruments and synthesizers?

[MindWanderer]

Aha! I keep hearing how everyone drops high-pass filters on everything every chance they get, but I never really understood why before. Thanks Zircon!

[timaeus222]

I' ve just checked out the EQ thing.

It lowers the at most possible master volume (right below clipping) a lot - just by ear I would says it makes a difference up to 30 % of loudness (if you put an EQ over every instrument or not).

That's the gain I would need to make it sound like the other songs.

@timaeus222: So you would everything try to do to avoid equalizers (or just the EQ presets?) in general and make the whole sound just by the instruments and synthesizers?

EQ is basically like chiseling out the imperfection in a sculpture.

I don't avoid using EQ because everyone needs it eventually, but I try to get the sound as good as possible before equalizing, sure. The better it is before you do anything to it, the less you need to do to make it sound good enough according to your own standards. Less work = faster work. Saves me time later on whenever I design my own patches, not to mention when you make your own stuff, you are familiar with how it can be tweaked and can make good use of it.

Edited June 20, 2014 by timaeus222

[Nabeel Ansari]

Here: http://downloads.izotope.com/guides/iZotope-Mixing-Guide-Principles-Tips-Techniques.pdf

This is a nice guide which explains to you not only how these tools work but why you use them. Having people tell you what EQ settings to use for certain things isn't really helpful if you're not really clear on why the equalizer exists in the first place.

Remember that technology comes as the result of necessity. The best way to learn how to use it is to use it only when you feel it's necessary. Don't ever do things just because they make the project more complex. Your mastering chain should be an input and a desired result, not an input and different sounding result.

Knowing when to use effects comes from practice and feedback. Example, once you learn what crowded mids sounds like due to people telling you your mids are crowded, then you know every time you hear crowded mids to fix the problem on your own before you show it to anyone.

Also, in this context, it is worth pointing out that mixing and mastering are indeed very different things, and you should learn the principles of a good mix before trying to master your stuff. Good mixing can survive without mastering, but mastering can not survive without good mixing. In the industry, the mastering engineer would send back a bad mix to the studio who mixed it, because mastering can't solve instrument balance issues and qualities of the individual elements of a track.

If you want general loudness advice, though, start with a good visualizer, like Voxengo SPAN. SPAN will tell you the RMS and the crest factor. RMS is basically the mathematical "perceived loudness", or the actual energy of the sound. Crest factor is just the mathematical ratio between the peak amplitude value and the peak RMS value. Crest factors have been getting really small these days (about 6 dB, which sounds kinda crunched and terrible), but for good loudness without in your face brickwalling, you should try to shoot for a crest factor around 10-12 dB. (once again, Crest Factor = Peak / RMS)

Edited June 22, 2014 by Neblix

[MindWanderer]

RMS is basically the mathematical "perceived loudness", or the actual energy of the sound. Crest factor is just the mathematical ratio between the peak amplitude value and the peak RMS value. Crest factors have been getting really small these days (about 8dB, which sounds kinda crunched and terrible), but for good loudness without in your face brickwalling, you should try to shoot for a crest factor around 12-14 dB. (once again, Crest Factor = Peak / RMS)

I'm still having trouble with this.

I noticed that in the mix I'm working on right now, the crest factor goes up when my mix gets louder. During the quiet buildup, crest factor is less than 1. Then it gets to 6-7 as I add instruments, and about 15-16 when I start hitting the limiter. That's with SPAN before the master limiter in my mix; when I move it after, the numbers change but the pattern remains (about 10 when I start limiting). I also notice that SPAN keeps track of the max crest, which suggests that big = bad.

I'm also pretty sure that crest, being a ratio, is unitless (not measured in dB.

On a related topic, I see the following two comments frequently:

1) You should set your master limiter at around -0.2dB.

2) TLs Pocket Limiter is the bomb.

Except, as near as I can tell, TLs doesn't let you set a limit at anything but 0dB. Are these two statements contradictory, then?

[Master Mi]

Nice EQ guide - thx.

I guess with the RMS thing and the crest factor I would handle it more practical to get the right loudness of the song.

Since I have another stereo headphone without an extra volume control (so a fixed volume of the hardware)- and after I have finished my first remix at the right volume level - I will remember the exactly level of the master volume where it sounds just right and full in things like loudness - and add a fixed size of dB in the master volume afterwards to get the right volume for the export version. =))

Monkey tactics I guess - sometimes old but gold.

[Nabeel Ansari]

I'm still having trouble with this.

I noticed that in the mix I'm working on right now, the crest factor goes up when my mix gets louder. During the quiet buildup, crest factor is less than 1. Then it gets to 6-7 as I add instruments, and about 15-16 when I start hitting the limiter. That's with SPAN before the master limiter in my mix; when I move it after, the numbers change but the pattern remains (about 10 when I start limiting). I also notice that SPAN keeps track of the max crest, which suggests that big = bad.

I'm also pretty sure that crest, being a ratio, is unitless (not measured in dB.

Crest factor is referred to in dB as a convention. Don't think about it too hard. ;_;

Keep SPAN after your limiter. Doesn't do you any good for mastering if you put it before. You want to analyze your result, not your input. (Though you can add a second one before for comparison)

Crest factor is the ratio of your peak to your RMS. It's not something you follow along every step of the way, you look at the over-arching picture of the song itself. Max crest factor is what you're going for, because unless you're doing something terribly wrong where your automating the dynamic range of your song with a gain-limit control, the max should tell you all you need. So again, I amended the numbers to be a little more realistic, you want to try to get around 10dB or above for your max crest factor. Anything less is gonna start sounding smushed or saturated (or both). RMS, I usually keep the max around -12dB (max meaning it's -12, so -11 and -10 go above the max).

On a related topic, I see the following two comments frequently:

1) You should set your master limiter at around -0.2dB.

2) TLs Pocket Limiter is the bomb.

Except, as near as I can tell, TLs doesn't let you set a limit at anything but 0dB. Are these two statements contradictory, then?

And... I'm not sure who's telling you to set your master limiter at -0.2dB. When you're mixing, you want headroom, yes, but...

Do you know why you're told to leave headroom? You want headroom because maximizing your music for printing and distribution is not something you're focused on in the mixing stage. That's mastering. You leave headroom when you're mixing so that the mastering process can then say "I have all this headroom, let's fill it up now". 0dBFS is the equivalent to a maximum floating point waveform (decimals ranging from -1.0 to .999...). This is why 0dBFS is the maximum for digital signal processing. Any values outside the range don't exhibit the behavior of numbers inside the range; if you multiply decimals within -1 to .999..., you get an absolute value smaller than your factors (.5 * .5 = .25), while outside of -1 to .999..., they get bigger (3 * 5 = 15). This is important in DSP techniques, trust me.

There's no formal reasoning for a limiter at -0.2 dBFS, afaik, other than habit by whoever's telling you. If you want to put TLs at -0.2, put a gain effect after TLs and lower it by .2.

Since I have another stereo headphone without an extra volume control (so a fixed volume of the hardware)- and after I have finished my first remix at the right volume level - I will remember the exactly level of the master volume where it sounds just right and full in things like loudness - and add a fixed size of dB in the master volume afterwards to get the right volume for the export version. =)

Can you explain what you mean? This doesn't quite make any sense; you're further adding dB to your master after you've determined it sounds full and loud? Why?

Edited June 22, 2014 by Neblix

[Master Mi]

Maybe I should have mentioned that the volume that I hear in my DAW project is far louder than the maximal volume I can hear in the Windows Media Player and after the upload on Youtube/Soundcloud after the exported version.

So the maximal volume level sinks a lot after exporting.

So I have to make the Master Volume a bit louder (about 30 dB) than my ears like - to achieve the same loudness in the exported version of my composition.

[timaeus222]

Uh... Did you turn off all the audio effects from your Sound options in the Control Panel? Try that if you haven't already.

[Master Mi]

You mean the FX/Equalizer panel up there in the mixer? (audio next to it is just the recording function for the live input of real instruments and voices)

http://magazine.magix.com/de/wp-content/uploads/sites/2/2013/04/open-vst-via-mixer.jpg

No I haven't done this yet in my newest uploads - but i've tried it once and it makes the master volume about 30 dB louder without FX/EQ settings - but the same goes for the exported version.

[Moseph]

There's no formal reasoning for a limiter at -0.2 dBFS, afaik, other than habit by whoever's telling you. If you want to put TLs at -0.2, put a gain effect after TLs and lower it by .2.

The technical rationale for limiting lower than 0 dBFS is that there are some situations in which a digital-to-analog converter can produce clipping-like errors when converting samples at 0 dBFS. It happens becauae the DAC has to generate a voltage based on the digital samples, and depending on the arc of the waveform, the analog peak may occur between two max-value samples, which causes it to exceed the digital value (this is called an inter-sample peak). You can actually get metering plug-ins that will tell you how high these peaks will be, but the simplest way to make sure they don't cause problems is to limit your master at slightly below 0 dBFS.

[Nabeel Ansari]

The technical rationale for limiting lower than 0 dBFS is that there are some situations in which a digital-to-analog converter can produce clipping-like errors when converting samples at 0 dBFS. It happens becauae the DAC has to generate a voltage based on the digital samples, and depending on the arc of the waveform, the analog peak may occur between two max-value samples, which causes it to exceed the digital value (this is called an inter-sample peak). You can actually get metering plug-ins that will tell you how high these peaks will be, but the simplest way to make sure they don't cause problems is to limit your master at slightly below 0 dBFS.

That makes sense. I was specifically pinpointing -0.2 (I don't see an issue with -0.1 in this regard).

Maybe I should have mentioned that the volume that I hear in my DAW project is far louder than the maximal volume I can hear in the Windows Media Player and after the upload on Youtube/Soundcloud after the exported version.

So the maximal volume level sinks a lot after exporting.

So I have to make the Master Volume a bit louder (about 30 dB) than my ears like - to achieve the same loudness in the exported version of my composition.

What DAW are you using? Also, I would suggest opening the Windows mixer and making sure all programs are outputting the same level.

Edited June 22, 2014 by Neblix

[Neifion]

That makes sense. I was specifically pinpointing -0.2 (I don't see an issue with -0.1 in this regard).

-0.2 to be absolutely safe, while being imperceptibly lower than 0.0 to the average listener. I've read books that say either -0.1 or -0.2, but I do -0.2 just to be safe.

[Master Mi]

@ Neblix:

I use Magix Music Maker - it's the lil bit lower priced version of Sequoia/Samplitude (which is pretty tough with about 2500 bucks (Euro)).

It guess it's not so famous at the global market (maybe) but I decided to take this one which has much more solid basis of functions and a lot more instruments and synthesizers compared to the standard version of Fruity Loops oder Cubase Elements which are nearly the same price section.

Just bought some of the extra instruments here (which are actually pretty good stuff) - but at the third of the download price by buying some very cheap previous versions of the Music Maker where these instruments have been already included.

http://www.magix.com/de/hidden-extras/zusatzinstrumente/

[timaeus222]

You mean the FX/Equalizer panel up there in the mixer? (audio next to it is just the recording function for the live input of real instruments and voices)

http://magazine.magix.com/de/wp-content/uploads/sites/2/2013/04/open-vst-via-mixer.jpg

No I haven't done this yet in my newest uploads - but i've tried it once and it makes the master volume about 30 dB louder without FX/EQ settings - but the same goes for the exported version.

The extra "effects" that are supposed to "improve" your listening experience. There is most definitely an option that lowers the volume of "loud" music to "help" you, but obviously it doesn't help music producers since it badly alters the final result. I just turn it all off so what I should hear after I export is exactly what I hear; what every producer wants to hear is exactly what he/she exports with no sound alterations.

It's in the Windows Control Panel in the Sound options, and I guarantee it's on by default on a brand new (factory settings) Windows computer. I'm not talking about Magix anything. Btw, 30 dB? o.o; Something at -30dB is about as audible as subtle violin harmonics, so something 30 dB louder is really loud. It takes maybe 5 dB for most people to hear a difference in perceived volume, IIRC.

Edited June 22, 2014 by timaeus222

[Nabeel Ansari]

Something at -30dB is about as audible as subtle violin harmonics, so something 30 dB louder is really loud. It takes maybe 5 dB for most people to hear a difference in perceived volume, IIRC.

3dB is a doubling of power; though generally with beginner ear training, yes, 5dB is barely distinguishable, being almost 4x the power. You have to know what you're referring to when you say difference, though. Are you talking about adjustments? Back to back comparisons?

In mixing you can hear down to a difference of a quarter of a dB when doing EQ stuff, but with back to back chunks (like two songs), it's a little harder because your brain can't distinguish power levels that small through memory.

Edited June 22, 2014 by Neblix

[timaeus222]

3dB is a doubling of power; though generally with beginner ear training, yes, 5dB is barely distinguishable, being almost 4x the power. You have to know what you're referring to when you say difference, though. Are you talking about adjustments? Back to back comparisons?

In mixing you can hear down to a difference of a quarter of a dB when doing EQ stuff, but with back to back chunks (like two songs), it's a little harder because your brain can't distinguish power levels that small through memory.

Yeah, I mean immediate volume adjustments (I didn't know about the 3dB = twice the volume thing; are you sure? Not 10dB, or are we on different scales?), but in terms of entire song comparisons, even different songs, I think that's what the website I saw this on was referring to. In EQ though, yes, at least for me, I've been able to hear differences in 0.4 dB occasionally, but it depends on the frequency you're boosting as you know.

Edited June 22, 2014 by timaeus222

[Nase]

In EQ though, yes, at least for me, I've been able to hear differences in 0.4 dB occasionally, but it depends on the frequency you're boosting as you know.

It's possible you heard the difference, it's also possible your mind kind of produced or exaggerated it for you, ya know?

Ever had one of these moments where you're fiddling with a knob and think you're doing some subtle but noticeable fine tuning, then realize it was the wrong knob, or the fx plugin was turned off or something?

Placebo effect in music production is kinda interesting, hehe.

But i'm just talking about my own experience, maybe your brain works a little more analytically. Pretty OT anyway.

[timaeus222]

It's possible you heard the difference, it's also possible your mind kind of produced or exaggerated it for you, ya know?

Ever had one of these moments where you're fiddling with a knob and think you're doing some subtle but noticeable fine tuning, then realize it was the wrong knob, or the fx plugin was turned off or something?

Placebo effect in music production is kinda interesting, hehe.

But i'm just talking about my own experience, maybe your brain works a little more analytically. Pretty OT anyway.

Yes, once, but I was actually thinking, "uh... why is this not working? D:", then I eventually realized I was tweaking the wrong automation clip (same color, different row =P). [/OT]

Edited June 23, 2014 by timaeus222