Hmm, hard concept for me to understand. I still don't get what is headroom. Is it the gap between a good volume level and when the sound starts to clip or distort? 

I think I'm going to keep my master bus meter below 0db. At the most 1db. Does that sound reasonable?

Yes.


If you analyze commercial CDs you'll find that peak levels can be anywhere from -12db to -0.0db, with the bulk of pop 'n rock records around -0.1 to -0.3db. If you send your music out for professional mastering it'll likely come back pushed right up to -0.1db. But -1.0db is a more realistic (and safer) practical target for us amateur MEs.


It is important to note that the subjective loudness of the recording has little to do with peak levels, but rather average levels. You use a limiter to make sure peaks don't exceed a specified target (e.g. -1.0db) and then start increasing the volume going in to the limiter. This raises the apparent volume of the music, at the expense of the crest factor (the ratio of average to peak levels). 


Go too far, and the music starts to sound dull and lifeless and lacks clarity and punch. Don't go far enough and the record sounds quiet compared to commercial releases. Mastering is a compromise between those extremes. Which way you lean depends on the genre and your own personal taste and style.

It's a huge topic, but here's a quick recipe - not necessarily the final word but just one way to get it done:

Insert an instance of Voxengo SPAN on your master bus as the last plugin in the chain, after the limiter. Set it to Mastering mode and select K-14 metering. Set your limiter's brickwall limit to -1.0db. 

Start by selecting a section of your song that represents the average volume of the song (e.g. a verse, not an intro, fadeout, or a big chorus) and play it back from there. Adjust the limiter's Input slider until SPAN's output meter is bouncing around the 0db mark. (If your limiter does not have an input slider, use the master bus Trim/Gain slider.)

Next, select the loudest passage in your song and play that back. Watch the meters, and if it jumps above +6db for more than an instant, pull back the limiter's Input slider until the song mostly stays between 0 and 6db.

That's it. This procedure will get you in the ballpark, and you'll have commercial volume without sacrificing too much clarity and dynamics. If your musical style wants to be louder, switch to the K-12 scale and follow the same procedure. If your musical style is, say, acoustic bluegrass or a chamber orchestra, go down to the K-20 scale. If your musical style is EDM or rap, well, feel free to ignore all of the above advice. 

From your explanation above compressor and limiter sounds about the same. Though I'm sure there is a difference.

I like your advice and will use it on my next project. Though I'm probably going to let it peak only at the 1db mark.  The reason is that of all the 3 CD's Iv'e imported into cakewalk, the song that had the highest peak was at +.6db only.  

Yes, a limiter is actually a type of compressor. Although there are several internal differences between a compressor and a limiter, the main functional difference is how stringently they apply compression. A limiter will allow you to specify a hard upper limit, a so-called "brick wall" that the limiter will not allow your data to exceed, no matter what.

You don't want to ever let your master bus exceed 0.0db, because even though that's acceptable within SONAR, it's not acceptable for an audio interface. Not just your audio interface, but also your listener's iPod or CD player. It's the limiter's job to make sure it doesn't happen.

So I hope you meant -1db, not 1db. :)

"Yes, a limiter is actually a type of compressor. Although there are several internal differences between a compressor and a limiter, the main functional difference is how stringently they apply compression. A limiter will allow you to specify a hard upper limit, a so-called "brick wall" that the limiter will not allow your data to exceed, no matter what. "

So does compression let the data pass the threshold? Also what do you mean by "data"? Level? "You don't want to ever let your master bus exceed 0.0db, because even though that's acceptable within SONAR, it's not acceptable for an audio interface. Not just your audio interface, but also your listener's iPod or CD player. It's the limiter's job to make sure it doesn't happen.  So I hope you meant -1db, not 1db. :) "




I'm a bit confused and most likely mixing the two, but when you said this:


Next, select the loudest passage in your song and play that back. Watch the meters, and if it jumps above +6db for more than an instant, pull back the limiter's Input slider until the song mostly stays between 0 and 6db. 


were you referring to to master bus? Also on a side note, I have seen recordings done with the master peaking at +.6db. 

If you imported a song from a CD and saw it peaking at +0.6db, then what you saw was an artifact of the conversion. A CD cannot physically store a +0.6db signal.

When I referred to keeping your signal (yes, on the master bus) below +6db, that's in the context of K-metering. A K-meter is calibrated differently than the normal meters you're accustomed to. When you use the K-14 scale, the 0db reference point is moved down to -14db, so hitting +6 actually means you're hitting -8db on the "real" scale.

Late post. (Nothing to see here)

"If you imported a song from a CD and saw it peaking at +0.6db, then what you saw was an artifact of the conversion. A CD cannot physically store a +0.6db signal." 

How would you accurately read the dB signal of a CD song? "When I referred to keeping your signal (yes, on the master bus) below +6db, that's in the context of K-metering. A K-meter is calibrated differently than the normal meters you're accustomed to. When you use the K-14 scale, the 0db reference point is moved down to -14db, so hitting +6 actually means you're hitting -8db on the "real" scale. "


I see... So if I get your point, I should add a limiter to my master bus to make sure my levels don't go past the 0db mark on the master bus meter. The recommended peak level on the master bus meter should be between the -14db and -8db mark?


Question is, what's wrong with using the master bus meter? why use the k-14 scale?

sharpdion23


"If you imported a song from a CD and saw it peaking at +0.6db, then what you saw was an artifact of the conversion. A CD cannot physically store a +0.6db signal." 

How would you accurately read the dB signal of a CD song? "When I referred to keeping your signal (yes, on the master bus) below +6db, that's in the context of K-metering. A K-meter is calibrated differently than the normal meters you're accustomed to. When you use the K-14 scale, the 0db reference point is moved down to -14db, so hitting +6 actually means you're hitting -8db on the "real" scale. "


I see... So if I get your point, I should add a limiter to my master bus to make sure my levels don't go past the 0db mark on the master bus meter. The recommended peak level on the master bus meter should be between the -14db and -8db mark?


Question is, what's wrong with using the master bus meter? why use the k-14 scale?

As Bit says - the reference point in the K-Metering system is moved. Such a system is a "perceived volume" scale. Just like Equalization is perceived volume, when it's really volume of a frequency. Now, this doesn't mean just because the K-Meter says -8dB that you should adjust the master to offset it. IOW, you don't want the perceived volume (as indicated by the Spectrum Analyzer) to reach 0.0.


Boost the bass frequency, that part seems louder than everything else. Boost the mids, instruments in that range are perceived to be louder than they really are. Same is true with the treble. Now, pull up a spectrum analyzer, and you'll see this being the case. Take a track and pull all the bass, mid and/or treble out. Those respective parts of the spectrum will decrease (ie: perceived volume) and the conventional meters will react in the same manner. Likewise, if you boost those parts. (I just pushed/pulled the mid/treble frequencies and this is what happened. To my ear, they got louder/softer in volume).


The conventional meters tend to react to bass boosts/reductions more often (and more vividly) then the other two frequency categories - you could squeeze mids in there, but low mids probably.

Going back to the K-Metering system: Now, this doesn't mean just because the K-Meter says -8dB that you should adjust the master to offset it. IOW, you don't want the perceived volume (as indicated by the Spectrum Analyzer) to reach 0.0. I currently have a track set to K-14. Depending on what "0" is on the VU meter, the reference point for the perceived level is -14; however, this doesn't mean I actually have more headroom. If I took this as just, I'd be beyond clipping!

To show you what the Spectrum Analyzer plug looks like: http://www.440audio.com/en/software/v4292-Voxengo-SPAN/#win

Regarding the Limiter, I can't speak for Bit, but given my experience once I found out what the Limiter was actually doing, I have avoided using it altogether. As I said if it's just turning up the volume, applying a Limiter should be the absolute last resort! (And last thing, in the chain if you do use it)


Bit, correct any mistake I may have made

I found this youtube video that explains k-metering. http://www.youtube.com/watch?v=GnREPzUfUgU

Let me know if there are something he said wrong.

"When I referred to keeping your signal (yes, on the master bus) below +6db, that's in the context of K-metering. A K-meter is calibrated differently than the normal meters you're accustomed to. When you use the K-14 scale, the 0db reference point is moved down to -14db, so hitting +6 actually means you're hitting -8db on the "real" scale. "


If in K-14 scale 0db is really -14db on the real scale...What about the k-20 and k-12 scales?