Off-the-record: ProAudioVault

ProAudioVault is Ernest Cholakis and Dan Dean. We loved their first release, a grand piano virtual instrument called Bluthner Digital Model One (or BDMO, for friends), and since they’re also among the most respected people in the software industry we wanted to ask them some questions about BDMO, sampling and related topics.
Well, it took months for Ernest Cholakis to get its homework done, but you know, developers are always super-busy trying to do the right thing for us, isn’t it? 😉

I know you and Dan come from different backgrounds, how did the idea of a collaboration come out?

Over the years Dan and I developed a friendship. Obviously sampling and recording technology is something that we both have in common. We produce different content – Dan has released mostly orchestral and bass samples libraries where I have released drones, drum loops, DNA groove templates and reverberation impulse products.
We often discussed various sampling issues, both technical and marketing as well as the growing problem of piracy. When we started to consider the Blüthner Piano sampling project we decided to pool our combined knowledge base because of the considerable technical challenges involved such as the signal processing required for noise reduction, recording procedures like how to record without any ambience etc. Because the piano is one of the most difficult instrument to record effectively, we decided our combined expertise could develop a better overall product than if either one of us took on the project alone.

Before starting working on BDMO which were your feelings about the available software pianos on the market?

We obviously thought that there was still room for improvement as most libraries had too much ambience in the piano samples and true piano dynamics were not accurately captured. In 2005 I wrote an article in Sound On Sound on this subject – (issue Nov DVD 2). THe article points out that all the major samplers could not accurately reproduce the piano’s dynamic range – even when the right tone was put into any given sampler, the dynamic range was inaccurate due to the sampler engine limitations. Most samples (even new releases) still cannot accurately reproduce the dynamic range of each individual note of a real world piano.

BDMO really sounds like a labour of love (just have a look at the manual and you’ll agree with me). How much time did it take?

Planning started months prior to the recording sessions which took place the end of October 2004. We released it April 2007 so about 2 1/2 years.

Why did you choose a Bluthner? And, how did people at Bluthner collaborate with you during the making of BDMO?

After playing the Bluthner for 5 minutes at Annehein NAMM in Jan. 2004 we knew that the Bluthner Model One was the piano to sample – it has a warm and distinctive tone that was never harsh – even if you play fff. The other element that appealed to both of us was the evenness of the tone. This is clearly apparent as one plays up and down the keyboard. An another reason for choosing this piano is that is has a distinctive sound that is different from the Steinway and Borsendorfer which already have been sampled. We met Dr. Christian Blüthner formally at Musik Messe Frankfurt the following April.
Ours was the first sampled piano library to be officially endorsed by the piano manufacturer. For both Dr. Christian Blüthner as well as us, the most important criteria was quality. As well we all agreed that recording at Skywalker was our first choose as it has one of the quietest soundstages in the world and for a piano this is an essential consideration.

This product, besides being an excellent sampled library, has many interesting additional features: the most important is the IR one, which really makes the BDMO unique. My only critic is: don’t you think this important feature could be better implemented in the GUI, to be more explicit and user-friendly? Users could easily get confused with those cryptic preset names, and for some things they need to keep the manual always handy…

Yes it is a point well taken. There are two reasons.
In the end one should listen to each timbral impulse get to know them individually then decide which one(s) work well in a composition. In a DAW a user can change the timbre without having to re-record.
Clearly RI users find it very appealing of obtain the impulse of a famous acoustic environment as well as a visual of the space. But in a musical sense why does this matter? I dealt with this same issue when I released both Pure Space Reverberation Impulse CD’s – I refused to mention and include pictures of the spaces because too many people out there are listening with their eyes more that their ears. The purpose in audio production is to find the right type of ambience and adjust it if need be to compliment or enhance the music. Whether the space is well known, one’s bathroom or a garage should be immaterial to the decision of selection. That is why I developed the bass, midrange,and treble numbers to describe the “character” of a space – because they actually make more sense in terms of describing a sound than a picture can ever do. In the end does anyone really think that if they have the Concertegedouw or Abbey Road RI’s that is is going to sell more records ?
Another reason I decided to withhold detailed information on the sources was for legal reasons.

BDMO recently got an important update/bugfix. Would you like to say something more about it?

Yes it was released in January 2008. We changed about 800 notes in subtle ways but with a distinctly noticeable improvement in the overall sound of the piano. We also added a script so that a user can adjust the keyboard sensitivity. This feature enables one to customize the response for any controller.

Ernest, judging from your background and the BDMO’s Timbral IR list, you have a library full of incredible recordings, probably made in every corner of the planet. I’m sure you have some interesting and funny anedocte about this, to share with our readers…

No funny situations I can think of but definitely a few surprises.
I find it especially interesting when one is in a space that has a historical significance. Over the years I have recorded in many temples, tombs, churches, grottos and early Christian churches. I remember when seeing the film of Pink Floyd Live at Pompeii the juxtaposition of this classical ambient band in a historical setting (Roman amphitheater) with no audience was poetic. So when traveling to Italy my partner and I when to Pompeii and recorded the RI of the amphitheater and man of the other buildings.
I find it fascinating that the sound of a particular space is not what one would expect. For example in the Pantheon one would think that the reverb time would be long and thick sounding given the incredible size of this temple however is is surprisingly short and depending where you stand has only a few discrete echos in the “ambience” of this space. Another space that I find interesting is Brunelleschi’s Pazzi Chapel in Florence. It is not a big space but the reverb time is longer than one would expect (maybe due to the architecture’s symmetry) and on top of that – it is one of my favorite places in Florence to visit!

Probably the biggest change we’ve seen in the ‘sampling world’ in these latest years has been the exponential growth of the libraries (thanks to larger and cheaper hds), which have become bigger and bigger. And sometimes this does not automatically means having a better and more playable product.
Now, in 2008, which is your vision on the state of art of sampling, and which are the next steps this technology is going to take?

Yes I agree bigger is not necessarily better. I think the large library trend has evolved for a number of reasons. Marketing, piracy issues and quality issues. All soundware developers are forced to deal with piracy and as internet and hard drives get larger so it seems the level of piracy also goes up. One very basic form of copy protection is large sound libraries. The problem with this approach is that given the level of polyphony required for a large ensemble such as a sampled orchestra a lot of pressure is put on the streaming engine of an sampler.
Earlier this year I took make a violin resonance impulse of an Andrea Amati and used it to convolve with a regular sawtooth wave patch. The realistic vibrato and character of the resulting tone was very dramatic to my ears. I can’t help but think synthesis will go down this path and that convolution will likely be integrated into more sound-ware libraries and software synthesizes.

I know it’s not your cup of tea, but still I’d be interested in your opinion…
Since some years we’ve seen a growth of audio modeling based products. Do you think there’s a future for this kind of technology, especially regarding the emulation of a complex instruments like a piano?

Obviously modeling will continue to handle ever more of the piano tone nuances as the processors get faster. A piano sound can be modeled but modeling different instruments (i.e. Blüthner versus Steinway versus Borsendorfer) is a challenge because each piano manufacture and even piano’s of the same model all have different partials (tuning and amplitude) and they all change differently as the sound evolves from ppp to fff. The amount of data required for a realistic sound is staggering especially if the user want much more than a “generic” grand piano sound.
That may take a while and will see if the marketplace even cares enough about this to support this kind of development.

is ProAudioVault just a ‘one product brand’, or are we going to see more products under the PAV name in the future? I swear I won’t tell anybody…

Who knows, maybe… (smiles)

Thanks Ernest and Dan!

By the way Ernest added some interesting news. He released a new interesting reverberation impulse CD called “Hollywood Impulse Responses” under his own brand, Numerical Sound. This what he told us about this CD: “These RI’s are unique and have something that no currently available software/hardware reverberation unit can do – add reverb that also changes the timbre of the source material as well as being designed for specific instruments (can’t be done with any EQ).” You can check some audio demos which illustrate how HIR dramatically changes the timbre of an audio track at Numerical Sound’s website.