Saturday, February 6, 2016

Interview: Mercuriall Audio

Mercuriall Audio is an interesting audio company from Russia, founded by the programmer Vladimir Titov and mix engineer Vjacheslav Tikhonov in 2015.
The company has started releasing some free plugin (that still today can be downloaded via their website), before making the leap into the realm of paid Vst Plugins, with their first paid creations, Amp One, a virtual amplifier for Ipad, and Tube Amp Ultra 530, a vst that simulates an Engl E530 Preamp (Click Here for our Review)
Here is our interview:




GuitarNerdingBlog: Hello and welcome to Guitar Nerding Blog! Can you give us some details on how Mercuriall audio was born?

Mercuriall
: Hello Atoragon and everybody who reads Guitar Nerding Blog! About 5 years ago I first tried to make a convolution reverb algorithm, starting from a simple (and very slow) method. 
And for my first program - guitar cabinet emulator Cab 1.0. I was looking for a simple name little in common with the cosmic theme. But that was not a real "birth" of the projectm it was a project that I was doing as a hobby in my spare time. 
At that moment, I experimented on modeling guitar cabinets for Free amp-sims. The real birth of Mercuriall can be considered the moment in which I created the preamp-sim JCM800. It happened when I was not working, therefore I was completely focused on modeling guitar cabinets and preamps. It was a very difficult time for me as I was working on free software and did not have sources of funding within six months. Then I started working on Amp ONE for the iPad, which took me about a year. After the release of Amp ONE, it only slightly eased my financial situation. Tube Amp Ultra 530 came about a year after the release of Amp ONE. U530 became my first commercial software for desktop computers. A more logical to assume that the release of U530 and the launch of the website Mercuriall.com is the official birth of Mercuriall Audio Software. In the beginning, I did not imagine that it would have been so difficult, because since the beginning of the work on the Amp ONE and before the U530 it took about two and a half years during which I had no funding and I worked completely alone at my own risk. It was a difficult time, which I would not like to repeat. Perhaps the birth of something wonderful always happens in terrible agony. In this case, it happened to delight fans of high-quality tube amplifiers modeling.
If to be completely honest, I haven't worked on the U530 alone. I entrusted the work on the circuits of E530 and GUI for new plugin to one of the Amp ONE beta testers. He as well as I plays guitar but also understands the circuitry better than I do. I needed someone who is willing to experiment with the circuits of the analog prototype, someone who can rewire or make measurements in the right places of circuits, to find places where there is a differences between an analog device and simulation. Together we succeeded to finalize the U530. Therefore Mercuriall Audio Software - more than one person, it's a small team.

GNB: What do you think about the amplification business nowadays?

M
: Honestly, I cannot say something very specific about today’s amplification business, as it is hard to make comparisons with the past. I think it is a very exciting time now, as the processing capabilities of modern computers allow us to do much more than it was possible before, and it is not only because of the hardware engineers’ achievements, but also because of programmers, who have learned to use hardware at its full potential.
At the same time, many large companies who create software for guitar players had too much focus on the number of simulated amplifiers and pedals, sacrificing the quality. Now they cannot move away from this path to create new products with just a few devices, but of the highest quality of simulation. They are forced to use marketing and redesign tactics to stay in the business.
This gives room to startups who focus on the most accurate modeling of amps and pedals. At the moment, all the necessary theoretical materials are available through the internet. Learn and create!

GNB: What is the design of a Mercuriall audio plugin usually like?

M: Previous free plugins were mostly experimental. I tried different approaches to achieve specific goals.
For example, working on U530 consisted of several stages.
Since we were modeling a tube preamp, the first thing we did was to compile the scheme of the device in LTSpice IV simulator and test its operation in comparison to the actual device.
At the next stage I moved the LTSpice-model directly to the code in C ++, so that the program could perform the simulation – but not in real-time, as the simulation algorithm requires complex calculations for solving systems of nonlinear differential equations. At this point, I ran the simulation and generated a response to a given stimulus to get the teaching sample set to train the neural network, which will replace the individual algorithm for solving nonlinear differential equations and will significantly accelerate the simulation of the circuit.
In parallel, we worked on the user interface of the program in the Blender 3D.
Afterwards, everything is put together, tested, adjusted and finalized.
The longest and hardest part is working on the visual appearance of the program.



GNB
: What is your opinion about the world of amp/cabinet simulators nowadays? Do you think eventually they will replace entirely the classic hardware?

M: Entirely – they will never be replaced, of course. We can draw an analogy between real and electronic books. Traditional books are still popular despite the fact that we have an opportunity to read them electronically. I think the main point here is that it is important for people to have something that is less dependent on “time” and that can be considered as a reference that is the base for the progress. Nevertheless, most will use the amp/cab-sims because it is more convenient.

GNB: What is the philosophy behind your software: analog modeling, black box approach or else?

M: I think all amp-sim developers use the principle of the black box. If you put a device with full analogue modeling (e.g. LTSpice IV) one the left side and on the right a device based fully on the black box principle (Kemper), then the majority of those will be somewhere closer to the black box. Our approach is very close to the version with full analogue modeling (LTSpice IV), because it uses the same methods for the simulation of the analog circuit as it happens in the simulation in LTSpice. The difference is that to speed up the calculations, some of them are «simulated" by the neural network, which essentially can be considered a black box algorithm integrated into the analog circuit simulation.

GNB: What are your career highlights as a software house?

M: (This is probably a long story) I first started to get acquainted with programming in 1994, when I was a schoolboy studying in the 8th grade, and studied at the courses Programming in Microsoft FoxPro for DOS. Then I did not have a computer and it was interesting and mysterious. 
I got a PC one year later, and three months after its appearance I had a software package Symantec Zortech C ++ 3.0. At that time it was the best compiler C/C ++ for DOS / Windows 3.11, and I started to learn C/C ++. Then I was in the 9th grade of high school. 
The problem was that I was not understanding what I do, and what I wanted to get from it. 
Back in '94 on the computer courses they showed us how to work with spreadsheets. I was amazed that you can enter a mathematical expression into a table cell, and instantly receive the calculation result, without compiling any of the code. 
I became interested in it and look how that information can be programmed. As a result, I found the book describes the basic operation of BASIC interpreter in C. And I had the idea to create a shell of their language, similar to Pascal to take part in the competition program in the 11th grade. At the competition, I won a thick book Internet Explorer 3.0 and a box of floppy and mouse pad. This is how my passion for programming started. My project has not received further development, but this was the first experience of creating programs with more than two thousand lines of code. While studying at the University I got interested in the modeling of neural networks, but since in the curriculum we did not have anything connected with the study of neural networks, it was mostly searching and reading on the subject material from the Internet. 
After the graduation, I have worked as a teacher of computer science, but the lack of prospect forced me to look for something else.  
I have worked for 6 months as a programmer in the Kamchatka center of communication and monitoring: there I was programming in Perl and Delphi (Pascal). From there I went to the company that provides access to the Internet (the company - provider of the Internet) where in addition to Perl, I had to write programs in PHP, PL / SQL (Oracle). There I was again faced with a C when it was necessary to modify the source code for Sendmail and Cyrrus. So once again I began to write code in C. Next came Mercuriall Cab 1.0 for the first time I took off impulses from my own cab and tried to apply machine learning to get orthogonal series Volterra instead of the usual IRs. And when the need for a programmer disappeared and I got fired, I became more focused to lead the development. The first amp which appeared after my dismissal was JCM800 that was absolutely free. Then came JCM800 Hot (JCM800 modified scheme), Harlequin. Then came the same plug-ins but with improved performance and accuracy of the modeling. Donations received was very small, but I was able to save enough money to buy iPad Air, and start developing a program Amp ONE for the iPad. It's the main stages at the moment.





GNB: How do you think the future of digital amp simulation will be? Will it be still for years a continuous modeling of old hardware or do you think someday the concept itself of amplification will change, along with the tastes of musicians?

M: Hard to predict. I think that the success of the accurate modeling of guitar amps will first lead to users wanting to get back to the origins of the guitar sound. Perhaps there will be experiments to obtain new sounds with the desired characteristics, but models of transistors/pentodes/transformers - these will remain as close to the originals as possible. Thus, the sound search will be carried out by changing the scheme of "virtual" amplifiers, and if desired, it will be possible to re-create the same amplifier in the analog world. In principle, this is how modern amplifiers are created - starting with the device’s circuit in the emulator of electric circuitries. At the same time, trying to match the sound to the “industry standards” of Clean, Crunch and Lead.

GNB: Do you have any tip or suggestion on how to use your plugins? Or some setting you prefer?

M: At the moment, I use default controls of the Free-version of U530. The reason is that I do not deal with professional mastering and mixing, and for playing in headphones, such settings are just enough.

GNB: Thank you for the interview, you can conclude by saying to our readers anything you want!

M: I want to thank all those who believe in us. We will continue creating wonderful things for you all. We have big plans for 2016. We will continue to work on the precise modeling of amplifiers. And right now we are working on our new project - Spark!




No comments:

Post a Comment