Reply to this topicStart new topic
> Mastering - How To Get A Loud 'commercial' Master, Part 2
Saoirse O'Shea
post Dec 17 2010, 01:47 PM
Post #1


Moderator - low level high stakes
Group Icon

Group: GMC Senior
Posts: 6.173
Joined: 27-June 07
From: Espania - Cadiz province
Member No.: 2.194



In this previous thread here I described what the recording artist, mix engineer and producer can do to help get a 'loud' recording.

What I now want to do is discuss what a mastering engineer may do. Some of this may need to get technical but I'll do my best to explain things as we go along.

-----------------------------------------------------

Caveat 1: Gainstage calibration.
I will assume that the DAW and all analogue hardware has been gain staged and calibrated correctly at 0dBVU. If they are not then changes in gain on the hardware may lead to unwanted clipping of the DAW and possible damage to your hardware.

Signal waveform and level
If you look at a waveform as displayed on your DAW's track lane the wave will look a little like a mountain range with some high mountains, some plateaus and some low valleys. The summit of the tallest mountain in the range is the maximum signal and is often referred to as the peak signal and has a value in terms of dBFS (decibel full scale); on a digital DAW the dBFS nominally should not normally exceed 0dB. If you were to measure the average height of all the mountain range you would have the the rms (root mean square) in dB. The difference between the rms and the peak value is the crest factor and gives some indication of the dynamic range of the audio material.

Lets take an example where the highest peak has a value of -3.1 dBFS, the RMS is -14.1 and there is a crest of 11. We can achieve a 3dB increase in gain by taking the signal up to -0.1dB, just below the nominal 0 threshold by lifting the fader. A 3dB gain will result in a 50% increase in apparent volume. Now the peak will be -0.1dBFS, the RMS will be -11.1 and the crest will again be 11. At this point the signal has increased in level and so will sound louder but we have not changed the profile of the waveform at all. (All we've done is pushed the entire mountain range up a few feet.)

If we decided that we want even more volume several things will happen. First peak signal reach the the nominal 0 dBFS threshold for a daw. Second the RMS will increase and the crest factor will reduce. Lets assume we add 4.1 dB of gain to our example. The highest peak is nominally 0dBFS, the RMS is now -10dB and the crest has reduced to 10. If you look at the waveform the tops of some of the highest peaks will have a flat top cut at the 0dB threshold and the dynamic profile of the wave will look a little squashed in comparison with the original.

If we keep on increasing the gain more and more peaks will reach the nominal threshold of zero and the profile will start to look less like a mountain range and more like a flat brick. At some point in this process the audio will lose most of its original dynamics and will start to sound flat and distorted.

The importance of crest to loudness
The crest factor for a lot of modern music the crest factor sits between 14 and 10. If you reduce the crest factor too much the music will lose it's natural dynamics and impact. In music some passages sound loud and have impact because they are louder then quiet passages. If all the passages were the same volume the music would lose impact.

Consider this, if you are in a quiet church and some one shouts something the shout will sound loud because of the large dynamic difference between the average rms volume in the church and the sudden peak of the shout. Think of the same thing happening in a busy, noisy bar. The background rms is much higher and there will much less dynamic range between this and the shout. In some bars you literally have to shout to be heard above the background noise: here the dynamic profile would look a little like a brick. The church has a large crest factor, the bar very little. As we keep increasing the gain of audio and reducing the crest the music will start to lose impact and will not increase in perceived loudness. Perceived loudness depends on dynamics. Shouting in a church sounds louder because of the difference in dynamic range.

How to get volume part a/ Clipping and what happens at the nominal 0dBFS threshold
It's generally believed that audio cannot exceed 0dB in digital equipment. This isn't actually the case: 0dBFS is a nominal threshold beyond which clipping and distortion may occur. Below 0dbFS the signal is passed to the DA converter which then reconstructs the original analogue waveform. At and above 0dBFS the signal is chopped off; the DAC now attempts to reconstruct the waveform based on the signal prior to 0dBFS. So the DAC recreates the wave above 0dBFS but the wave will be a modified, distorted version of the original, real signal. As more signal exceeds 0dBFS, more will be clipped and the more the DAC has to reconstruct but based on decreasing amounts of original signal information. At some point if enough signal is clipped the DAC's reconstruction will produce excessive, distorted white noise or digital distortion.

Why is this important? First, reconstruction means that a peak signal can exceed the nominal 0dBFS threshold. Second, that at some point too many cumulative peaks will result in digital distortion. Third, it may be possible to allow a small number of isolated transient peaks to exceed 0dBFS without noticeable distortion (distortion is still there, it is just not obvious) because the DAC reconstructs the transient relatively well and because the transient passes by before we register it fully as audio signal. For this to work the peaks here must have a very fast attack and release and little sustain: the peak rises very quickly to maximum and falls away very quickly i.e. it will only work with some (not many) types of instrument and some (not many) recordings.

Based on this a mastering engineer may decide that an audio file can be allowed to deliberately clip in order to drive up the peak signal beyond 0dBFS and so raise the RMS whilst maintaining crest. This is sometimes referred to as clipping the ADC output and is done by deliberately sending a hot signal from the output of the analogue to digital converter in to a DAW (NB caveat 1). To do this the equipment must be accurately calibrated and gain staged and the ME needs to listen very carefully for distortion in the output signal.

How to get volume part b/ Editing
Lets return to our mountain range. Assume that most of out mountain peaks are roughly a height of 10k meters and two are at 15k. Those two very high mountains will reach the nominal threshold much sooner than all the other mountain peaks. Much the same happens with a lot of audio waveforms, the majority of peaks sit a few dB below the highest one or two. A good human drummer will produce most of their transients hits at @ the same level but a very small number will be noticeably louder. (A mediocre drummer will produce inconsistent transients and the volume will go up and down randomly.) We can edit the very highest and reduce them to @ the same peak height as the majority of the others. What we have done is sacrifice some crest for a lower but more consistent peak signal. This lower signal in turn will allow us to now increase the peak to the nominal 0dBFS and increase the RMS without further reducing the crest.

Caveat 2:
Don't edit all the transients to the same level as it will then sound mechanical. Some dynamic variation makes the transients sound more human and more musical. If you use a drum vsti program you may want to consider adding a little dynamic variation.

How to get volume part c/ EQ
The audible range for human hearing runs from 20Hz to 20kHz. However, humans do not perceive these frequencies equally. We hear some frequencies more clearly and easily than others: we hear the high mid range of 1-4kHz much more readily then sub bass below 60Hz. A consequence of this is that the low bass and sub-bass parts of the frequency spectrum require a lot of energy to be heard as loud as the high mid. If we can reduce the low bass and sub-bass of an audio stream without it having a detrimental affect then we will free up a lot of potential energy that may be used to increase the volume of the audio overall. This may be achieved by high pass filtering at, or shelfing down, the very low frequency range. A HPF may for instance be used set at -6dB per octave from 45 Hz to slowly reduce all the frequencies below 45 Hz. It is however very dependent on the musical genre and the instruments used in the recording.

How to get volume part d/ compression
A compressor may be used to deliberately reduce the dynamic range of an audio file and so reduce the 'height' of the highest transient and increase the RMS of the signal overall. This is similar in effect to editing the waveform. Compression values depend on the actual audio signal but very often in mastering compression ratios may lie between 1 and 3 to 1 ( i.e. 1.1:1 through to 3:1) with a threshold value of less than -40dB for most of the signal but the highest peaks may have additional, and much more severe, gain reduction. This means that the compression apply is minimal but applies over most of the signal and only the highest peaks are then deliberately heavily compressed or soft clipped: most of the audio passes through virtually unchanged but the top peaks are deliberately taken down a few dB. After this the signal can be increased in volume without exceeding 0dBFS.


How to get volume part e/ Limiting
Clipping occurs when the signal exceeds the nominal 0dBFS. We can avoid, or at least minimise the chance of clipping if we place a limiter at the end of the signal chain. A limiter is essentially a compressor that is set at a high compression ratio (ratios here are from 10:1 to infinite:1). A brickwall limiter is a limiter that is also set to prevent any peak exceeding it's set nominal maximum. (On a normal limiter it is still possible for a peak to exceed the nominal maximum.) As a signal approaches the limiter's set maximum the limiter will start to function. The more gain reduction we apply the sooner the limiter will function and clamp down on the signal. So we can set a limiter to clamp down on a lot of the signal or hardly any. Theoretically the more the limiter clamps down the more we can increase the rms. This however may come at the expense of crest. In addition high levels of gain reduction may result in very noticeable and unwanted audio effects. The signal may start to sound as if it's pumping as the limiter noticeably kicks in and out. Additionally the limiter may have a sound of it's own which becomes more and more obvious as gain reduction increases. One way round this is to use several (2, maybe 3) limiters in series to form a chain. Each can then be set to achieve much less gain reduction individually whilst the chain itself still achieves a large amount.

Caveat 3:
Intersample peak overs. As already mentioned a DAC will attempt to reconstruct any signal that meets or exceeds 0dBFS. Several sequential, fast transients can produce an effect called intersampling overs even if the peaks individually do not meet or exceed 0dBFS. Here some dacs and cd players will struggle to reconstruct the waveform and if this occurs they may either skip or stop playing or produce white noise.

Clipping without limiting may leave audio open to intersample overs. If a limiter is used however it may be set so that the output is limited to a given maximum value and from experience an output value of -0.3dbFS will achieve this.

How to get volume part f/ which technique to use
Any and all. What you use to achieve volume in mastering depends on the material. Nonetheless most of us mastering engineers would prefer not to have to slam music just to make it 'loud'. Music sounds better with natural dynamics. Personally I try and retain a reasonable crest value and only reduce it if a client instructors me to. Even then there are some instances where I will inform a client that it is against my professional experience to do so.

One recent example was where we mastered a film soundtrack. The producer asked for it to be as loud as a recent commercial CD she liked. Our response was that EBU standards require -12db RMS and no clipping and that could not be achieved at the same time as the commercial volume she wanted. She insisted so we did as she asked and the first master was rejected by the company. It was accepted once we re-mastered to the EBU standard.

In conclusion, if you want your CD loud my best advice is to retain crest and use the volume control on your stereo.










--------------------
Get your music professionally mastered by anl AES registered Mastering Engineer. Contact me for Audio Mastering Services and Advice and visit our website www.miromastering.com

Be friends on facebook with us here.

We use professional, mastering grade hardware in our mastering studo. Our hardware includes:
Cranesong Avocet II Monitor Controller, Dangerous Music Liasion Insert Hardware Router, ATC SCM Pro Monitors, Lavry Black DA11, Prism Orpheus ADC/DAC, Gyratec Gyraf XIV Parallel Passive Mastering EQ, Great River MAQ 2NV Mastering EQ, Kush Clariphonic Parallel EQ Shelf, Maselec MLA-2 Mastering Compressor, API 2500 Mastering Compressor, Eventide Eclipse Reverb/Echo.
Go to the top of the page
 
+Quote Post
OzRob
post Dec 17 2010, 03:19 PM
Post #2


GMC:er
*

Group: Members
Posts: 468
Joined: 26-September 08
From: QLD, Australia
Member No.: 5.967



More great reading. Thanks Tony, I'll have to return and absorb this when I'm not so tired. BTW, I've been playing wth Voxengo Elephant and Flux Limiter demos and A/B/C-ing them against my current Crysonic Spectraphy. To me, they honestly all sound as good as each other. Dunno if that's the case or if my ear is just not attuned enough to pick out the differences.


--------------------
------------------------
My website (Flash) My other website (non-Flash)
------------------------
Go to the top of the page
 
+Quote Post
Saoirse O'Shea
post Dec 17 2010, 03:36 PM
Post #3


Moderator - low level high stakes
Group Icon

Group: GMC Senior
Posts: 6.173
Joined: 27-June 07
From: Espania - Cadiz province
Member No.: 2.194



Don't know about the Crysonic Rob as I've never used it.

To me the Elephant is very transparent even if you push it hard with lots of gain reduction but it can sound a little too cold and characterless on some material particularly if you oversample. The Flux is not quite as transparent but I think it has a nice colour to it - almost analogue like the hardware Manley Slam but much cleaner and with a much better noise floor. I tend to use the two in series if I need a fair of gain reduction to get the level but with a little warmth.


--------------------
Get your music professionally mastered by anl AES registered Mastering Engineer. Contact me for Audio Mastering Services and Advice and visit our website www.miromastering.com

Be friends on facebook with us here.

We use professional, mastering grade hardware in our mastering studo. Our hardware includes:
Cranesong Avocet II Monitor Controller, Dangerous Music Liasion Insert Hardware Router, ATC SCM Pro Monitors, Lavry Black DA11, Prism Orpheus ADC/DAC, Gyratec Gyraf XIV Parallel Passive Mastering EQ, Great River MAQ 2NV Mastering EQ, Kush Clariphonic Parallel EQ Shelf, Maselec MLA-2 Mastering Compressor, API 2500 Mastering Compressor, Eventide Eclipse Reverb/Echo.
Go to the top of the page
 
+Quote Post
OzRob
post Dec 18 2010, 12:59 AM
Post #4


GMC:er
*

Group: Members
Posts: 468
Joined: 26-September 08
From: QLD, Australia
Member No.: 5.967



That's interesting about the Elephant in your experience. When listening, I like the Flux slightly more but couldn't say why. Maybe that's it.

If you ever want to demo Spectraphy you can check it out here. I don't know about all the technicalities and marketing spiel, but to me, it sounds good, however I've never needed to push it or brickwall limit anything. I have tested it with all settings on extreme and there is pumping and distortion but I don't know when one would ever need it set like that, unless it's for experimental shaping or something.

In fact, if you want to pick it up let me know. As a user, they sent me a no-brainer price. I can give it to you. It's not an affiliiate link or anything, just useless to me as I have their whole collection. Only downside is it expires on the 21st Dec.



--------------------
------------------------
My website (Flash) My other website (non-Flash)
------------------------
Go to the top of the page
 
+Quote Post
Todd Simpson
post Dec 18 2010, 02:36 AM
Post #5


GMC:er
Group Icon

Group: GMC Instructor
Posts: 15.404
Joined: 23-December 09
From: Atlanta, Georgia, USA
Member No.: 8.794



Some very good advice here as usual. Well Done. Always pro tips and tricks. The "Volume Wars" seem to continue unabated. Sadly, many seem to have let go of the concept of "Dynamics" in favor of having as loud a master as possible. Clearly you are not promoting this which is good to see. Things run in cycles and I can't wait until the "as loud as we can master it" cycle runs it's course.


--------------------
Go to the top of the page
 
+Quote Post
Saoirse O'Shea
post Dec 18 2010, 12:32 PM
Post #6


Moderator - low level high stakes
Group Icon

Group: GMC Senior
Posts: 6.173
Joined: 27-June 07
From: Espania - Cadiz province
Member No.: 2.194



QUOTE (OzRob @ Dec 18 2010, 12:59 AM) *
...
If you ever want to demo Spectraphy you can check it out here. I don't know about all the technicalities and marketing spiel, but to me, it sounds good, however I've never needed to push it or brickwall limit anything. I have tested it with all settings on extreme and there is pumping and distortion but I don't know when one would ever need it set like that, unless it's for experimental shaping or something.

In fact, if you want to pick it up let me know. As a user, they sent me a no-brainer price. I can give it to you. It's not an affiliiate link or anything, just useless to me as I have their whole collection. Only downside is it expires on the 21st Dec.


Thanks for the offer Rob B) - very kind of you although we don't need another limiter at the moment smile.gif .

Pumping can work for some types of audio - some D&B tracks and as you said some experimental work well with some pumping. I was playing one of Lamb's early cds a few days ago and there some compressor pumping used on that well. Distortion - white noise etc can work but only if you can fully control it and use it how you want for affect. (Can't remember the title of it but there are some very effect bursts of white noise done for effect in both a Matmos track and Wolf Eyes.)


--------------------
Get your music professionally mastered by anl AES registered Mastering Engineer. Contact me for Audio Mastering Services and Advice and visit our website www.miromastering.com

Be friends on facebook with us here.

We use professional, mastering grade hardware in our mastering studo. Our hardware includes:
Cranesong Avocet II Monitor Controller, Dangerous Music Liasion Insert Hardware Router, ATC SCM Pro Monitors, Lavry Black DA11, Prism Orpheus ADC/DAC, Gyratec Gyraf XIV Parallel Passive Mastering EQ, Great River MAQ 2NV Mastering EQ, Kush Clariphonic Parallel EQ Shelf, Maselec MLA-2 Mastering Compressor, API 2500 Mastering Compressor, Eventide Eclipse Reverb/Echo.
Go to the top of the page
 
+Quote Post
Saoirse O'Shea
post Dec 19 2010, 02:20 PM
Post #7


Moderator - low level high stakes
Group Icon

Group: GMC Senior
Posts: 6.173
Joined: 27-June 07
From: Espania - Cadiz province
Member No.: 2.194



QUOTE (Todd Simpson @ Dec 18 2010, 02:36 AM) *
... The "Volume Wars" seem to continue unabated. Sadly, many seem to have let go of the concept of "Dynamics" in favor of having as loud a master as possible. Clearly you are not promoting this which is good to see. Things run in cycles and I can't wait until the "as loud as we can master it" cycle runs it's course.


Absolutely Todd smile.gif ,

We signed up to this campaign some time ago. Record label people in the industry are starting to get the message.


--------------------
Get your music professionally mastered by anl AES registered Mastering Engineer. Contact me for Audio Mastering Services and Advice and visit our website www.miromastering.com

Be friends on facebook with us here.

We use professional, mastering grade hardware in our mastering studo. Our hardware includes:
Cranesong Avocet II Monitor Controller, Dangerous Music Liasion Insert Hardware Router, ATC SCM Pro Monitors, Lavry Black DA11, Prism Orpheus ADC/DAC, Gyratec Gyraf XIV Parallel Passive Mastering EQ, Great River MAQ 2NV Mastering EQ, Kush Clariphonic Parallel EQ Shelf, Maselec MLA-2 Mastering Compressor, API 2500 Mastering Compressor, Eventide Eclipse Reverb/Echo.
Go to the top of the page
 
+Quote Post
Ivan Milenkovic
post Dec 28 2010, 04:01 AM
Post #8


Instructor
Group Icon

Group: GMC Instructor
Posts: 25.396
Joined: 20-November 07
From: Belgrade, Serbia
Member No.: 3.341



Excellent series of articles, always interesting to read Tony. Thank You! smile.gif


--------------------
- Ivan's Video Chat Lesson Notes HERE
- Check out my GMC Profile and Lessons
- (Please subscribe to my) YouTube Official Channel
- Let's be connected through ! Facebook! :)
Go to the top of the page
 
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 


RSS Lo-Fi Version Time is now: 16th December 2017 - 06:01 PM