CD Sound Master Mastering Engineer Michael Angel Angel Lofte Studio

~~www.CDSoundMaster.com~~

"Maximizing The Sound Of Your Recordings" TM

(Pictures, Before/After Demos, Mastering Tutorials, and More Coming Soon)
__________________________________________________________

{Home} {About Me} {About The Mastering Process} {My Mastering Philosophy}
{Mastering Tutorials} {Mixing Tutorials} {Links}
{CDSoundMaster Original Loop and Sample Collections}
__________________________________________________________

{AboutNebula} {My Nebula Libraries} {Mastering With Nebula} {Mixing With Nebula}
{Audio Comparisons}

__________________________________________________________

Mastering Tutorials

My goal for this section of my website is to provide practical tips and tricks for those interested in learning about audio mastering as an art and a science.

Mastering With Nebula
About Convolution
Nebula Background
Examples of using Nebula for Mastering

Mastering with Nebula

I am starting this tutorial with a specific program by Acusticaudio called Nebula, because the technology is so new and unique that it serves as an excellent example of a true "best of both worlds" in using high end hardware and using plugins for working with audio files. Historically, mastering has required very expensive outboard hardware in order to edit recordings without noticeable degredation, using digital sound as a means of preservation and working in a clean, unobtrusive environment. Until recently, it seems that most software has been secondary in doing the tasks involved in mastering except for the indexing and final CD creation process.

With a newly developed program called "Nebula", by Acustica Audio (Acusticaudio.net), the differences between digital and analog are coming together very quickly to bring a new way to master with all the benefits of both worlds. To appreciate what this program is working to achieve, it is helpful to look at what has led up to the current technology.

It has long been said that mixing "inside the box", or within the confines of the computer, could simply not compare to having the traditional hardware used in mixing and mastering. The division has been that the hardware is typically very expensive, and has a history of making the recording studio and mastering house an elite-club. For years, the home recordist has seen the prices come down on mixing consoles, recording devices, and consumer-level hardware, to a point where just about anyone had a level at which they could record themselves with a very good quality.

As digital technology continued to bcome more and more affordable, and computers got faster and faster, the market for bringing all the elements of a studio together into software programs became a huge new industry. From sequencing to notation, multi-track recording and mixing, to mastering, and even video editing, the modern studio can practically be carried around on a laptop with few limitations.

The draw back to the seemingly ideal digital studio, is that it is very hard for even the finest digital equipment to compete with the quality, musicality, character, and natural sound of high end analog gear. Where digital processing can excel at precise measurements, non-destructive editing, and very linear changes to sound, analog equipment has very complex elements of sound processing that tends to have more pleasant side effects than negative ones. Plugins that emulate such hardware have gained a great popularity, some doing a better job than others, and very few offerings that really restore what is done best with analog processing.

Return To Top

A Convoluted Theory

Several new technologies that have been improving on the process of emulating high end equipment are centered around what is called "convolution". The convolution process involves sending a test tone signal through a source sound, and processing the outcome of that signal. The resulting file is called an Impulse Response, or IR. It has been popularized by capturing the essence of live acoustic environments, making for incredibly detailed reverberation completely within the digital realm. Convolution can also take a virtual snapshot of frequency changes as they relate to the original volume of a signal, making it possible to gather a very basic, but often very realistic digital recreation of hardware equalizers.

The convolution process has been broadened into what is called dynamic convolution, where many settings are sampled and pieced together similar to snapshots being threaded together to make a moving picture. This technology has made for very convincing emulations of high quality equalizers, but also takes a great deal of processing power, and is protected by a patent that obstructs sharing the technology.

There are many benefits to this technology, among them being the use and popularity of digital 'samples' of very expensive gear that were unattainable to many home-recordists and hobbyists until the advent of the technology. However, many limitations found in the technology have continued to limit the consideration of widely adopting the technology into practice for mastering. Details of the interaction of phase coherency, the actual speed of dynamic information in impulse samples, filtering and deconvolving processes, not to mention the quality of source-sampling and equipment used, are all issues that may not hinder individual tracks in a mix as much as a final stereo mix of complex, multi-instrument program material. In some aspects, the elimination of certain distortion and other harmonic content from equalization curves can be a benefit, if properly aligned at very precisely aligned settings. But, even the slightest changes from sampling the original devices can make impulses unusable for a finished mix.

Return To Top

Nebula Background

Vectorial Volterra Kernels Technology
(Pizza, Mandolins, and the Genius of Simplicity)

Then comes "Nebula". The creator of the Nebula engine is Giancarlo del Sordo. His initial efforts were born out of the desire to eliminate some of the negative artefacts created by current convolution technology. Along with a few friends, Giancarlo was able to develop some results that seemed to maintain the integrity of the orignal signal being sampled much more succesfully than what had been done before. Giancarlo, along with the help of Francesco, Graziano, Antonello, and Mirco began to gather together to form Acustica Audio, enlisting a community of Beta Testers for debugging, testing, suggesting, and adding to the hardware being sampled. According to Giancarlo, the essence of a determination to do what had not been done, was a quality indicative of the Italian spirit. In seeking solutions to very complex issues dealing with improving on the concepts of convolution and dynamic convolution and developing a sampling process different from both concepts, the Acustica Audio team were able to create a processor capable of dealing with these issues in a brilliantly simple, elegant manner.

The current capabilities of this program reflect a major step forward in using digital processes to simulate those of it's analog sibling. By creating streams of information about the source's harmonic distortion, dynamic changes, and frequency curves in real time, "Nebula" is able to recreate a very real digital impression of analog hardware in real time. Where convolution can create a very precise single picture of a devices setting, Nebula is able to sample streams of information working side by side, and with enough resolution to be very convincing. Beyond the realism of the process lies other benefits, in that the unwanted side effects of the hardware being sampled can be altered and optimized, thus bringing us lower noise floors, increased dynamic range, and even the ability to gain one feature at the absence of another. We can use a tape machine in "clean mode" without imparting the analog harmonic distortion the signal would carry. We can process the benefits of the same device with harmonic distortion but without a change in frequency curves. The benefits of this technology are simply stunning.

Return To Top

Heavenly E.Q.

There are a couple of presets inside Nebula in particular that I want to focus on for use in mastering. The first three presets are called 2055 Angels EQ LO S, MD S, and HI S. These three presets divide frequencies into lows, mediums, and highs. The "S" in the titles stands for a technology recently added to the Nebula called Stone, Stone2, and Stone3, where adjustments can be made to optimize the original gear to make it more useful for mastering. Where there was loss in the high end frequencies of this original hardware eq represented, the "S" version maintains the benefits of it's character, while maintaining all the detail in the high end.

These presets are my first choice recommendation of eq's to use for mastering inside the Nebula, because they have enough of a pleasing character to be able to easily recognize what adjustments are being made, but they do not impart so much personality that the changes in boosting or cutting overtly change the character of the original material. It is important to distinguish these presets from other software 'mastering' eq's. There are many high quality software eq's available for use in mixing and in mastering, and some do involve the use of certain forms of convolution technology. Often, these eq's are excellent minimum phase or linear phase eq's. They are not geared towards sounding specifically like a certain device, but rather in solving the complexity of good sounding audio changes scientifically, with the fewest audible side effects. The presets I recommend inside the Nebula are not chosen becuase of their remarkable preciseness, although they work very well for a wide range of mastering tasks. They do mimic many of the pleasant characteristics of the original equipment in dealing with sonic adjustments with minimal side effects, and in a pleasing musical way.

When mastering, it is important to consider the best tool for the situation at hand. When making precise, careful changes you do not want to add or subtract any greater amount than is necessary to make the adjustment. If there is a spike in a certain frequency at certain portions of the music, you will normally want an eq that can isolate the width and volume of the spike, and reduce it in relation to the curve of the rest of the signal.

For example, let's say that the mix you are listening to sounds good, is very full sounding and has a pleasant rnage of dynamics, but there are intermittent sections of music where only a very specific frequency in the mid range of the music seems to jump out in such a way that it distracts from the energy of the rest of the recording. You don't want to change what is working, and you don't want to isolate one instrument as it relates to another. What I recommend in this specific situation is to place a simple brickwall limiter as the last plugin in your effects chain. Opening the Nebula to the "2055 Angels EQ MD S" preset, you have controls for gain, frequency, and "Q".

The "Q" represents the width of a frequency band and affects how much of the signal is going to be adjusted. The wider the "Q", the wider the range of frequencies will be adjusted. A very narrow "Q" will isolate a thin frequency band, and can be helpful for isolating problem spikes without affecting the frequencies around it. The original hardware represented by these presets is known for handling frequencies in a very natural way, imparting very little noticeable change to unwanted changes in the music, and leaving the timing of the sound well intact. There are several other excellent programs in the Nebula that have similar control over frequencies, but some of these are specifically designed to be used in the mixing process, where elements of desirable sound help enhance instruments and vocals in a pleasing way.

The "2055 Angels EQ MD S" has a "Q" range of .3 to 3.0, which is very narrow when pulling the fader all the way down, and very wide when adjusting the fader all the way up. For this example, lower the fader all the way down, and gradually increase the gain fader upwards. You will notice a specific frequency begins to get louder and louder. Now, take the gain fader and bring it all the way down. When it begins to move beyond the center point, it is at first difficult to hear the frequency going away, but after listening back and forth you will notice the changes more. The goal here is to find the offending frequency and reduce it to a level where it sits closer to the rest of the mid range material. The brickwall limiter is being used to prevent your volume from going beyond digital zero, protecting your ears and speakers!

What I recommend doing is bringing your Nebula gain up with the narrowest "Q" setting, and raise your frequency fader up or down, sweeping slowly through the frequency spectrum until you hear the offending frequency spike jump out at you. It should be obvious when you have found the right one, as it should be pronounced and louder than any others. Sometimes there are several frequencies, and other eq's are needed to best deal with them. In this case, once we hear the spike, we want to slide the frequency fader slightly beyond it upwards and downwads to get a feel for how much louder it is to the surrounding frequencies. If it seems to average about 5 decibels louder than the neighboring frequencies, then you can reduce it's volume from 1-5 decibels, depending on what you think sounds best. Keep in mind that it takes some learning and listening to get comfortable with this process, but once you have it down, you can easily discern how wide a "Q" seems right for the situation, and what is enough volume reduction to put an end to the spike. There are many factors involved in knowing how much to reduce in volume. If the frequency usually sits at a good level, only jumping out on occassion, then you are best in only reducing it by a decibel or two, as you don't want to take unecessary energy away from the good sounding sections of the song. You can also automate this volume change if it serves the song best. A third option involves limiting the frequency, but that is for another tutorial! When using transparent eq for mastering in this scenario, it is good to think of a narrow "Q" cut as an average of the offending peak's presence and when it is not present.

A narrow "Q" setting can be useful in other boost/cut scenarios as well. Let's say that there is an acoustic guitar part that sounds a bit dull and lifeless compared to other parts of a song's instrumentation. You don't want to adjust a lot of information in the upper mids, or cut too much in the lower mids, because in mastering you hope to preserve what has been recorded for the most part. Most likely, a narrow "Q" boost can center in on a particular frequency band that highlights a brightness, string overtone, or picking quality that is not as prevalent on other instruments. In this case, a slight boost at the right volume can enhance this performance without being an obvious change to the overall material. If there is too much conflicting information where you would like to boost the frequency, then another option is to center in on the low-mid frequencies that are a little more dull and lifeless, widen the "Q" just slightly, and reduce the gain by just a decibel or so. You should notice that the balance of what remains in the mix tends to work with the acoustic guitar in a more flattering way.

Having worked with many different eq devices and plugins, the ability of an eq to represent a narrow band of frequencies accurately varies greatly, when boosting at high volume so as to sweep through the spectrum to identify problem frequencies. I have come to prefer minimum phase, and musical sounding eq's even for a precise task like this, as I believe they represent the true sound better than more scientific, linear eq's. Even the best filtering process tends to add a false layer of edginess or harshness to the signal that was not there originally, when replication a very loud boost in a narrow signal band. A more musical eq is built to reproduce the signal truthfully, but moreso the way we perceive sound: more like a human hears it. If a seemingly precise eq represents a small boost of sound the same, linearly, at a louder volume, it no longer sounds the same to us, as this sound was never meant to be heard out of it's context with the rest of the music. Therefore, it can be tough to discern what to do when we hear this. The 2055 Angels EQ S handles this process very well, and will give you the information that you need to hear for sweeping through the frequencies to isolate problems.

Another benefit of this particular mid range preset in the Nebula plugin, is that it remains very musical and reasonably transparent at wider "Q" settings, even when using it to boost a signal. Let's say that you want to bring a vocal performance slightly more forward in the mix, and this is the lead vocal of the song. You can take the "Q" width somewhere to 1.0 and even wider, and listen to what range of frequencies accents the voice over any instrumentaion above and below it's constant frequencies. Amazingly, just a decibel or so of the right frequency band can make a dramatic musical statement, while simultaneously not announcing itself. The thing to look for in selecting the right frequency range, and the right eq for this purpose, is whether you are making an adjustment that you want to sound purely like the performance is actually slightly louder than other parts of the music, or whether you are wanting to add the effect of there being slightly more 'edge', or 'warmth', or other character. In this isntance, especially in mastering, an eq like this preset is useful in making the wide "Q" boost truly accentuate the performance in the recording without changing the feel of the recording sonically.

Return To Top

The Bell Curve and The Shelf

The "2055 Angels HI S' and "LO S' presets take on a slightly more advanced control feature than the mid range preset. The "Q" fader now doubles as an adjustment for the width of the frequency range at a setting of zero, and when brought all the way up to a setting of 100, it is now adjusted to working like a shelving eq. To understand the range of control available on these presets, you now have to learn two types of frequency filters, the bell curve eq and the shelf eq.

In the previous examples of working with the mid range equalizer of the 2055 Angels EQ S, we were dealing with what is called a bell curve eq. This type of eq centers in on the selected frequency, and boosts or cuts that specific frequency, with a slope slightly above and below that frequency which is determined by the width of the curve, or "Q". Now, we are introduced to a very high quality shelving eq in the low and hi frequency presets. The shelf style of eq will boost or cut all frequencies either above, or below, the frequency that is chosen. For example, with a shelf eq on a low frequency eq, we may set our frequency to 100 Hz, and when reducing the signal by 3 decibels, everything from 100Hz and below will now be cut by 3 decibels, with a slight sloping deeper as the frequencies are lower and lower. Just the opposite is true with the high shelf eq, where lowering a set frequency of 10kHz, for instance, by 3 decibels, will decrease all frequencies starting at 10kHz and above. Shelving eq is commonly used when mixing tracks to help eliminate the build-up of frequencies above and below the primary content being recorded on individual tracks. If one is trying to accentuate a bass drum, and the lower frequencies of a bass guitar are conflicting with the impact of the bass drum, a shelving eq can be placed on the bass guitar track, and the goal is to find what the frequency range is that is conflicting, and reduce the volume from that frequency down, the amount of gain reduction depending on what sounds best for the mix.

In mastering, it can be useful to reduce a wide slope of low or high frequencies, or boost the same. With a quality eq like the ones represented here, it can make very natural sounding adjustments to set the eq this way. Where you may need to eliminate specific problems with very narrow "Q" on bell curve style equalization, more sweetening can be done by shaping larger parts of the spectrum with the shelving eq. Be careful to listen to before and after settings several times before committing to any major changes. The more subtle and high quality this style of eq, the easier it is to perceive big changes as being smaller than they truly are, and until you are comfortable with the effect on program material, it is safer to make small changes.

Overlapping wide "Q" or shelving eq's like this which have gain adjustments, you can overlap eq's to mold the response of your recording in very creative ways. Making very small, overlapping changes can give your personal touch to a recording without making changes that detract from the intended sound of a mix. If there is a dull, lifeless high end, but the upper mid frequencies don't respond well to your various attempts at eq, you may consider bringing the frequency of the high shelf eq down as low as the upper mid frequencies, and lowering the gain by a very small amount. Then, with a second instance of the eq, experiment with raising the gain by slightly more than the first instance is cutting, and sweep through your high frequencies, both with a wide "Q" and with shelving eq, until you find a nice match that sweetens the previously dull high end, while leaving the natural feel of the upper mids unaffected, and yet providing space between the two for both groups of frequencies to breathe.

A Vintage EQ For Modern Recordings

The Vintage EQ S is another preset in the Nebula program that is just as useful in mastering as it is for mixing individual tracks. There are eq's that lean towards clarity and an almost clinical approach to modifying sound, and there are other devices that have more character, and are desirable for specific sounds and instruments. The equipment that producers and engineers prefer for different tasks largely helps to shape the sound they gravitate towards, and what clients hope to achieve by relying on their craftmanship in these choices.

Where a linear phase equalizer may be the best choice for trying to correct a problem in a recording without giving any indication that there was a change made, either in the sound, phase, or stereo correlation, sometimes (even in mastering) it is necessary to alter the sound of a recording to the benefit of the recording. Many eq designs provide stable, high quality adjustments with very minimal artefacts like distortion or phasing changes. Sometimes, what I call "musical" or "character" eq's are perfect for blending the feel of one song to better balance others on an album. I consider the Vintage EQ S of the Nebula to be of the calibre of these eq's. There is a consistent sound set to this preset, but it is not a perfectly precise device. The "S" setting quarantees accuracy in the upper frequencies beyond the original characteristis of the eq sampled. The "Q" values are adjustable, from a wide frequency range at a zero value, to a narrow "Q" at a value of 100. Boosting and cutting frequencies obtains a wonderfully musical result in many genres and styles with this setting, but it is always a good idea to test any changes against the original recording. Sometimes the variances that are sonically pleasing can be more difficult to discern, especially after longer sessions. Always make sure that the changes you make are not affecting other frequencies in an audibly noticeable manner.

Return To Top