Behind the Kernel (#9 - September 2012)

Officially Licensed 3rd Party Developer Libraries
Free 3rd Party Programs
User avatar
Posts: 2858
Joined: Sun Mar 28, 2010 9:00 pm
Location: Lodi | Madrid | Buenos Aires

Behind the Kernel (#9 - September 2012)

Post by enriquesilveti » Sun Sep 02, 2012 6:49 pm

Michael Angel from CDSoundMaster interviewed

Acustica Audio: How did you know about Acustica Audio and Nebula?

Michael Angel: Several years ago I was working on developing a signal-chain based concept for creating plug-ins that mimicked the behavior of analog gear better than the typical algorithmic approaches. I met Giancarlo on a forum that involved the topic of static impulse responses and he told me about the early stages of Nebula. I read more about the approach he was using and was interested in learning more, and we began emailing back and forth. I became a Nebula Free user, then Nebula2, and then began testing my signal chain ideas with NAT, and realized the true potential of VVKT. The thing that interested me the most in Nebula at that time was Giancarlo's intellect, kindness, and his open approach to accomplishing his goal.

Acustica Audio: Which feature you look into gear before get sampled?

Michael Angel: I am attracted to the elegance of great designs. Of course, it is always fun to find something rare and something extremely expensive to work with, but there are gems at every price level, and I tend to lean towards items that I know can not be done as well using other technology. Items like Pultec inspired designs that are amazing sounding, but no two are identical, and the unique qualities of every transformer winding and op-amp. I like things that have a subtle texture from harmonics, and this can be from a very small amount of non-linearity or more driven.
In the digital world, every plug-in can be valuable to bring the personality of the outside world into play. This is where something antique or textured or rough, in the right way, can be useful just as much as a costly eq. Having the right tools to add color, personality, and tonal texture to tracks in a mix can go a long way to making the mix and mastering processes easier, and then it is helpful to have the real analog response of mastering gear. Once again, Nebula is ideal for this! Essentially, if it has personality, always adds something unique, is not likely to sound as good using an algorithm, is very hard to find, and would belong in an equipment museum, then I am interested in it!

Acustica Audio: How you handle the noise or wow/flutter during the sampling process?

Michael Angel: Noise is an easier thing to overcome than wow and flutter, and anything that is purely timing based or phase related. If a device has excessive noise, then I do a series of routines to know what Nebula is responding correctly to and where it may stray. The difference between a noise floor and a clean signal produces a different kind of harmonic response than the range of a signal from very clean, usually right around unity, to driven or overly driven. I always test the dynamic range of the equipment and center around the cleanest signal, and decide the number of steps plus and minus based on single sample tests so I know that the harmonics will be adding the right information.
With Nebula, harmonics are always going to repeat an additive response, meaning that it is going to measure the difference from a very noisy signal to a clean signal with added harmonics, and if it is too noisy, it can start to accumulate bad information, so once I know the range where it responds accurately, then it becomes about the perfect combination of levels from the converters and the device.
Designs like op-amps, tubes, transistors, transformers, etc, are easier to reproduce using the VVKT process than tape, in that their timing can be calculated easier because they remain more linear. Where there is unique behavior and extreme fine details in how volume, spectrum, and harmonics change, they do not fluctuate constantly in time, so every layer of sampling tends to overlap without issue. With tape, you still want to have a multi-layered dynamic process and not a single sample, but every time those reels go around there is a potential for stretching, shifting, etc. Even with perfect transport and new tape, the smallest changes in pitch and tape stretch make it so that samples do not overlap correctly. This can be compared to the traditional 4-color process that we used to use in professional graphics printing. Each color had its layer, Cyan, Magenta, Yellow, and Black. There was a piece of film for these along with a marker to align these layers. We have all seen magazines that are not perfectly lines up, with that small 'plus' sign with misaligned characters that result in edging colors.
For more tricky signals like wow and flutter, I had a great deal of technical experience with this before working with Nebula, since I had already been working on research to recreate individual pathways ans signal chains to reproduce things like crossover distortion, harmonics, transformers, tubes in the preamp chain, rectification, etc. Sampling in Nebula becomes very challenging and for this reason, the basic decision was that tape was best to avoid. So, it took a long time to create a system for capturing tape accurately. It is something that was very involved and as a result I have more brands of tape media and tape machines than I would have ever imagined, and as a result I definitely hope that users can appreciate that R2R is something special. And, yes there are other things in the works that make use of the wonderful technique that makes these non-linearities a success. I won't share all secrets on this topic, as I would recommend NAT for other purposes that yield easier results more accurately for those looking to experiment with the process. I hope bringing items like R2R and TB+ and those that come in the near future document, archive, and preserve important pieces of analog history, and make accurate tools for more people than any other time or technology.

Acustica Audio: Which are the main differences from sample a single compressor or equalizer from a complete console path like libraries from Classic Consoles?

Michael Angel: I would have to say that the most distinguishing difference between a single compressor and an eq or console is that every aspect of the eq and console can be reproduced, with careful and accurate editing to a level that is almost impossible to tell from the real device, where compressors usually have a limited set of parameters that can be replicated to this degree. Also of note is that Nebula acts more as a replay button for things like eq and complex harmonic details like consoles, where the Neb engine itself is actually doing some of the timing element when recreating compression.
A typical single hardware compressor has a billion different potential combinations, and the difference between a signal that is being compressed with extremely fast attack and heavy threshold and ratio is vastly different than a signal that is getting just a small touch of timing change. Even the most linear analog compressor tends to have some slope and change to its timing, so if your attack is 10ms and release is 70ms and ratio is 2:1 and threshold -5, you may get variance if it is fed a sharp transient or drastic change from one type of signal to another.
Nebula handles specific complex details extremely well, but when it comes to a single device that can do a million different things that create very different results, the only way to catch every single details accurately is to add more and more kernels to cover every nuance. The best way to simplify the process is to reduce the number of variations in a single program, which immediately changes the flexibility of the device, but increases its benefits in use by loading quicker and acting more predictably.
The Nebula engine does more to enhance and reproduce the complexity of changes to timing in compressor programs, where it dedicates very specific kernel chunks to very specific data when operating on something like an eq or console. These can all be very complex sound processes, but something like a console is more predictable over time, and is mostly about doing the right tests, similar to the one's mentioned above dealing with noise and harmonics. Consoles are meant to be dynamic programs, meaning that they represent how the device acts at low volumes, medium volumes, and driven fairly hard. Items like TB+ and Tube Booster are designed to let the user add more saturation and gain, even to noticeable distortion, to signal that remain more subtle like tubes, consoles, and tape. Using separate instances of Nebula make it easier to build up the natural effect of harmonics on the signal.
Eq's are probably the most exhausting devices to sample, and truly require attention and patience. There are hundreds and hundreds of settings to record, test, re-record, and edit. Every setting has its own slightly unique harmonic structure, and sampling an eq is not as simple as changing dials over and over, but also means knowing exactly the best place to start because you are possibly going to be adding as much as 20+dB to your average signal, and taking away as much as -20dB. That is a huge dynamic range for something that is still set at a nominal average level before changing boost/cut.

Acustica Audio: Will be a next instrument compressors libraries for other instruments like vocals or electric bass?

Michael Angel: Yes. I am happy to say that I will be releasing a series of compressors for specific instruments and uses. This is already in progress and just takes a long time to reproduce and edit. Some items were produced at the same time as THE Drum Compressor, as I intentionally left some settings for processes that I like for bass, guitar, vocal, etc. Other items are still in experimentation. The goal with each is to bring to Nebula what we need from analog, and also to do something that you won't find anywhere else, yet has that wonderful analog vibe.

Acustica Audio: How hard is develop an Acqua FX from a Nebula library?

Michael Angel: It is hard. The process makes sense and it works, but I find it is typically just as hard for me each time I begin the process new. The foundation is the technology, and this remains consistent inside and outside Nebula, and still the majority of users are working with the Nebula host, but the stand alone Acqua's are beneficial in education people about VVKT and Acustica outside our wonderful community, and sometimes we work more naturally and fluid when we recognize the device for what it is. Still, it is a very lengthy process with new tests, lots of editing, and the challenge of answering technical procedures for two OS platforms, different bit processing, a dozen DAW hosts, and tricks that are unique to every single item above. It means knowing more than I ever wanted to know about Win7, Lion, etc.

Acustica Audio: why VTM-M2 does not use VVKT and how those 3 stage works all together?

Michael Angel: The VTM-M2 is designed for two main purposes. It is the realization of my dream to develop the full signal chain process that results in much more complex analog behavior than complex algo-based schematic design or table design, or process-oriented design, in this case used specifically to perform extremely complex limiting and compressing behavior of tape while matching the harmonic saturation character at the same time. I always try to explain the process like this: I am a believer in Nebula and VVKT. What it does, it does better than any technology, and no matter how the market changes over the years, the classic's inside Nebula will sound as amazing 30 years from now as they do today. But, as with any technology, there are some strengths and weaknesses. Nebula does not resolve the need for extremely high gain distortion, which is fine 99% of the time for consoles, tape, tube drive, etc. For high end gear of this nature we usually don't want it to go into extreme drive anyway. Nebula does not do extremely subtle levels of linear compression and it does not do limiting. Essentially, with limiting we want to be able to make extremely small, precise control over the entire signal that is unique to any other kind of reduction system. There are good algo's out there, but they always tend to reveal their 'sound' over time.
So, looking at the entire system of what tape does and how to replicate this, I studied the market and how it handles this. There are usually very generalized devices that do not specify a single machine, meaning that 'tape' sound does not mean a specific machine, but a tape-like saturation, or a tape-like 15 IPS or 30 IPS result, which is fine. If you look at a timeline, you will see that there are not any individualized tape machine plug-ins until after R2R. I strongly believe we pointed the market in that direction when they saw R2R's success. Sadly, VTM has recently been copied in name by competition, but all I can say is that there is only one real VTM!
R2R handles the specificity of individual machines, meaning that a Studer sounds different than a Wollensak. the spectrum responds different, as does the harmonic structure. So, you get the non-linear behavior of exact real machines this way. Tape Booster Plus uses single instances of Nebula devoted specifically to adding more natural tape harmonics to any choice of R2R machine. It is not necessary, but if you are going to use any tape 'saturator', then you are better off with TB+ for increased gain than other options.
The VTM-M2 can work on its own to give fully compressing, saturating, limiting tape response for tracks and for mastering. On its own, it can characterize the true essence of tape from very subtle to extreme. It combines the limiting and compressing element that Nebula does not do, but is carefully designed to fit perfectly into place after R2R and TB+ so that you have a perfect tape replication system. The essence of individual machines, the ease of using Neb to increase volume with saturation more realistic than others, and VTM-M2 to perform all limiting and compression character. To change machines, simply trade out your R2R and keep the rest of the chain in place. And again, you only purchase the element that means the most to your work. All three guarantees every aspect of behavior, and purchasing Nebula Pro, R2R, TB+, and VTM-M2 will still come out to a small expense than a single machine on the 'other' platform.

Acustica Audio: How much change the sampling method from Vintage Tube Collection to Tube Booster?

Michael Angel: The sampling process for Tube Booster is actually very different than the VTC. They were both a challenge, but for the VTC the issue started with something that was happening outside of the Nebula world. As a mastering engineer, I have some processes that I tend to use in certain situations. I prefer to lean towards maintaining the timing of a recording unless there are specific reasons to adapt the timing. So, when I want to get a touch of analog sound, I don't necessarily want to run a master through a compressor or limiter to get the tube element or transformers. there are significant parts of the devices that change the sound that I don't want there. So, I was searching high and low for a device that would be ideal for doing a final mix 'in a vacuum' so to speak, and I kept turning up results that were to the extreme. There were good sounding, but extremely cheap devices that were closer to the design goal, but still had other elements in the path and these elements were mass-produced so they couldn't be as picky or precise in how they were implemented. There were extremely expensive units that had one element of tube design or another, that I liked, but most of the cost went into elements that added more electronics than I wanted in the path. Truly, what I was after was more in line with hi-fi audiophile reproduction equipment. This is a crowd that may compare to the amount of detailed, focused listening that I do over the years of experiments. We study the sonic of natural sound reproduction to the same degree, most likely. But, I wanted a device that allows total flexibility and control over the amount of signal driving the tube, the biasing of the tube, and I wanted to be able to balance each chamber individually.
Now that I had a full appreciation of the rarity of such a device, I knew that I would want to provide this in Nebula form as well. It required successfully building the hardware device that would deliver the natural tube sound, and I think it does so well. The Source Plus became my dream for mastering directly through a perfectly tunable tube with nothing else in the path. So, then I searched the world over to find the most important and beautiful and rare tubes for the Nebula collection, and then worked for weeks on tuning the device the best way for Neb to interpret. And I later repeated this process again at 96kHz, as people had shifted from using 44.1kHz to 96kHz within those 3 years that passed.
Tube Booster presented a totally different challenge, which is to identify and quantify the most outstanding and unique harmonic behavior of the vacuum tube so that it could be added to any VTC tube and provide the same benefit as TB+ for tape. So, I thought in terms of this: when someone is thinking about the sound of tubes and wanting to record something through a tube, what are the qualities we are looking for. There is a different kind of drive than tape and something unique from any other element. So, I studied inside and outside the box, using several different tube devices and reproducing certain elements from ear until I felt that I knew which harmonics reproduced the right results. But, Nebula is also very particular with what is required to get the kind of harmonics that make the signal boost levels. This was partly in editing the way that the input and output correlate to the harmonics and partly going back to the drawing board several hundreds of time until the combination was right. Tube Booster was a very exhausting and often frustrating process. I had thought that the years of testing with tape would help me to isolate 'the' tube sound easier and quicker, but after a couple of weeks I learned my lesson. The result is something that I treasure and I hope it gets a lot of use. It is a little more subtle than TB+, but so are tubes. They tend to remain more subtle up to a greater amount of drive, and this layer and texture allows some pretty extreme harmonic injection before distortion becomes an issue. It also means that use after guitar pedals and amps can make for some of the most dimensional and amazing sounding guitars inside the computer- much better than any algo alone.

Acustica Audio: Can we consider The Source Plus as the hardware father of Vintage Tube Collection?

Michael Angel: Yes! Without the Souce Plus, we would not hear the tube as it exists on its own, but we would also hear everything else in the path. VTC would not be possible without succeeding at the hardware first. We are not currently building any new Source Plus', so it has become a very rare device. I hope to design more hardware concepts in the future. We already had one device completely finished and prototyped as well, but often people change paths or become busy in other projects and things don't happen that we wish.

Acustica Audio: Which studio configuration do you use. ITB, OTB or Hybrid?

Michael Angel: I use a hybrid configuration. I prefer increasing the signal with harmonics more over limiting and compressing, and other times a combination of both. If the sound I want involves more natural 'drive' than a single instance of Nebula, or more than just very minimal limiting, I will sometimes use my analog outboard gear. I have a few choices for mastering compression and analog preamps that I use to push the signal to puff things up a bit, but honestly I usually do not need more than what I do in Nebula. For eq, I almost never leave ITB. I have most of the originals right here to plug into when I need them, but I have the Nebula versions and they sound identical. So, I have the added bonus of seeing the hardware while listening inside the box. Nebula has truly altered my need for outboard gear for eq and most tape and saturation, but I still find it best to use natural limiting, and the Source Plus' can go into higher gain and extreme conditions if needed.

Acustica Audio: Do you use external DSP processing?

Michael Angel: I have two popular processors, one that operates on a PCI bus and one that operates on firewire. These have become more of a comparison platform than something I use often, but I am actually very impressed with the low to moderate levels of compression from the firewire design, the 'liquid' one. The "U" is a great tool for what it does, but there is not a single example of an eq that Nebula doesn't knock both options out of the water.

Acustica Audio: Do you think that audio engineering is better today?

Michael Angel: No. I wish it were, but I think we are experiencing something that is a parallel to the history of great societies. When we become so empowered by an over-abundance of what we need, we become a little less driven to work as hard to get the best results.
I think that we have the best potential for the best audio engineering today, but I think that the traditional role of training and hands on experience is not as good as it used to be. The skill of understanding the analog process used to help the engineer know how to get sounds, because we had some ideas of what produced those sounds. Knowing how to calibrate tape and how to literally set levels that give sonic results taught us so much about how to make sounds, where today there is instant technology for cheap or free, without any training or required learning curve. I like the potential for unexpected results, but I also wish that it were more common to treat the tools as the artist's palette, or the mechanic's finely chosen tool set.
I think that immediate access to hundreds of plug-ins at any given time is such a different world than knowing what makes the original hardware so unique. I think Nebula is a great resolution to this in some ways, as you have access to the people that work hands on with the analog devices, and you can reach actual real people to learn things. But, people are also swimming in a pool of corporate hype out there, and they are targeted for every single next best thing. I remember when the first 'mastering' plug-ins were introduced in the plug-in market, and there were all of these 'presets' that sounded terrible. I thought of it like having 5 options for auto-pilot for driving instead of learning how to drive. Or, do-it-yourself brain surgery; just press a button to make the best incision. The quality of most designs has matured, but the users still has little foundational understanding of the process to start with, so the potential is greater than any other time, but the average user starts with less of a need to learn a skill and perfect it, and a potential for great laziness and lack of imagination.
The other element that supports this is the unending world of visualization. In the analog world, we had to rely on our ears, finely tuned rooms, and excellence in reproduction. We had the choice of led's and VU's, and they were used almost exclusively for sound levels only. We could see if we were hitting tape low, medium, or high gain and we could input to unity on each channel, etc. But, we didn't have frequency visualization on every track which you can pull up instantly in your DAW. Even for mastering, we had desks with eq's and we may have an led to let us know we are overloading, but that is it. The ear is designed for an immense ability to discern micro-changes to time, spectrum, and volume. Having a little bit of visualization for the mix or mastering process was enough to produce remarkable sounding results. These days, people can spend way too much time looking at what they should be listening to.
to make things worse, there is no training in the difference between digital and analog metering, so people latch on to different fads in engineering that are largely meaningless, but completely alter the way that plug-ins have to be designed. Not only do they have to grasp how things would be in analog, but how they translate to digital. One of the things that becomes surprising is just how much fluctuation and nuance there was to analog gear, even extremely expensive 'linear' mastering gear. Much of that 'magic' analog sound is not nearly as linear as digital, and if we use our ears to discern the right uses for things, we learn much more about what sounds good than worrying over perfect linearity. If we never want any spectral or harmonics variance, then we are better with nothing but a transparent DAW and no inserts. But, sound has taught us that the complexity in analog non-linearity makes for better sound.
I recommend everyone to take the time to learn the roots of recording and learn methods before panicking about what the large manufacturers tell them they must have. Trends will come and go, and corporations will take advantage of these just to sell you something. Let your ears and talent dictate what you wish to achieve with your recordings and talents to make it a trade and an art form.

Acustica Audio: Did you have formal study in audio engineering?

Michael Angel: I had some formal training, a lot of hands on training, and a general interest in the field that made me a sort of electronics investigator. I had a natural interest in electronics at an early age, and would dissect radios, try to build radio shack kits, take apart realistic microphones, etc. I sang and played instruments starting at an early age, and had formal training in voice and percussion and played and toured some professionally. So, my interest in the recording field was there for many years. I listened to records not just for the songs as a writer, but analyzing the uniqueness of the sound achieved and I wanted to know how to do that. I first learned by reading the magazines, gear reviews, recommendations for how to mix this instrument and that.
I was in a band and we had access to a great live PA Our guitarist was the song of a locally famous restaurant/club that had a massive set of JBL and crown amps that we were allowed to permanently borrow for gigs. We had no idea what we were doing, and we ran these through the guitarist's Fostex 4 track tape recorder/mixer to have the inputs to mix from. I bought a Peavey mixer that sounded terrible, and started realizing we were never going to sound good as a band until we figured out how to amplify correctly. This process of writing, playing, performing, and doing a lot of things wrong led to my desire to get better. I made demos with the same Fostex that our guitarist left in my basement until surprisingly these started to sound good. I bought a Tascam for the same purpose, started researching and purchasing good mid-range gear, had my first decent console, and we started getting our recorded sound together.
We invested in a 3 song demo at a local studio and met a couple of aspiring sound engineers that we had the same taste in music and interest in gear. I started to learn some things about mic placement, levels, tuning, eq, expensive FX processing, and we sounded truly good in a recording for the first time. I remember the sensation was something like 'now I actually know what we sound like as a band!'
I continued to learn on my own and study all the trade magazines and built a small demo studio. Our guitarist went on to Full Sail, and we talked about our parallel experiences all the time. so, I learned through conversation what he hoped to learn hands on. As it turned out, he wasn't getting any hands-on time with the S*L's and N**e's as he hoped, but I was getting hand's on with all of the mid-range gear for 12 hours every day in my demo studio.
Over the years, I've read stacks of magazines on mixing, mastering, gear reviews, done hands-on time in a lot of studios, from professional small facilities to major commercial rooms. The hands-on time with engineers and mixers has always been the most valuable to growth, learning, and inspiration. I picked up a lot of theory from my 3rd person Full Sail classes via my friend/guitarist, and learned some fun tips and tricks this way, but 90% of it has come from doing and asking questions about what I did wrong.
I finally started to get really good at recording and mixing, had two DAT's for mixdown, and started recording all of the local bands and songwriters and word spread. I really didn't have the personality of wanting to be the local producer guy; I wanted to make my own albums, but I learned what I needed to know in working with others. As I made contacts with others that engineered and mixed, I continued to learn things. I recorded some tracks in a local facility that had a great A*I console and 16 track 2" and I learned from their sound guy every time I recorded.
When our band completed our first album, I researched all of the directories for the best mastering engineer and became good friends with Barry Diament, who in my opinion is still one of the most gifted sets of ears I've known. We wrote and talked about the science and art of recording and mastering, and not only did he master the first project but I ended up having the wonderful experience of learning under him. To this day, the things I learned from talking with Barry remain more true and consistently valuable than anywhere else I picked up information. Interestingly, the handful of my favorite articles ended up being written by him as well. Sound on Sound, Mix, and others had articles from Barry, and I didn't make the connection until later just how much I was already learning from him. To master projects that didn't have any budget to speak of, we used a local contact that became a close friend. This was when a good CD-R cost upwards of $500 and discs had come down from $12 to $8, yes for a single disc!
I saw the early days of crappy digital turn into the state of the art field it is today. I went from demoing in '89 to doing my first serious mastering.

Thank you Michael Angel!