Just my two sense,
I record british esque pop folk music with the piano being a primary force in most of my compositions. I\'ve just recently invested in the GigaStudio platform because I see it as having a great deal of promise...Seeing as I am a college student, I can\'t afford to spend a couple grand on an old steinway upright or comparable substitute...I LOVE MY PIANO SOUND DIRTY...I was so relieved when I heard someone mention the Abbey Road Steinway...that...in my honest opinion...is what I consider to be a truely wonderful sounding instrument..the reason...its organic...it has flaws...it has idiosyncracies that make it stand out...what i don\'t understand...and maybe you sample developers can handle this...why does everyone want their piano\'s sooooo clean??? meaning...its one thing if you are recording classical music...but thing of other genres....take jazz or pop....all of the great albums...their piano\'s are dirty...the sound warm and UNperfect...that...to me is what is so memorable about them...think back to the good ole days of the mellotron...those sounds have stuck around because the have that warm dirty sound...partially because its analog tape...but the point is....clean music isn\'t pleasing or beneficial to music...granted this is just my opinion....but most of the contemporary artists sound like ****....BECAUSE....their sound is too over the top...its too polished and clean...give us some noise....give us a rattle...these aspects paint the tonal landscape...they give us a sense of place within the music...clean music...clean samples for that matter...make us seem like we\'re in a sterile lab somewhere...there\'s no depth...no emotion to it...I am currently looking for the right piano for my recently acquired GigaStudio..and to be honest...the one that I am truely convinced of..as being the most...organic..truely playable sample is...the Malmsjo...it sounds warm and beautiful...maybe its just me...but i wish samples were less pristine...could some developer PLEASE provide those interested in producing quality ORGANIC music with a product that meets the needs...this applies to other instruments as well....namely...DRUMS....why can\'t a developer produce drums that sound like they just came off a Ramsey Lewis Trio session??? analog tape people....digital is too clean....give us the tape compression and warmth that only TAPE can give us...sure you\'re going to lose some in the A/D conversion...but still....conclusion...the warmth and idiosyncracies that are inherent in imperfect tape(or equivalent)recordings...gives the listen those subtle details needed to be transformed by the music...to be transplated...to be taken to the smokey club where jazz musicians thrive...to be taken to the hall where the concert pianists thrive...these little noises and attributes give the sound its three dimensionality...
Sorry for less the precise continuity...hopefully someone can relate to what im trying to get across!
Developers...keep up the good work...we\'re gonna make it!!!
By the way....Hans Adamson...you\'ll be hearing from me soon...LOVE THAT SOUND!!!
Personaly, I like my samples clean cause after that i can dirty them up any wich way i want. You want analog ? how about you take the clean samples and run em through a 2 inch... there\'s your analog.If you don\'t have access to a 2 inch use a tape saturation effect there are some really good ones out there. Want noise ? well play em through your monitors and record with a couple room mics at 22khz really low recording level and then normalize you\'ll get mucho mucho noise. (just kidding )
Happy Thanksgiving, really.. i was just kidding hehe
Hey JP, I agree with your sentiment that instrument imperfections can add a lot of character to a recording.
But there are a couple of important things for any library developer to consider:
1. If a piano disk is your first project, are you going to cater for a niche within an already small market - ie those who have a good standard piano and want something a little \'ragged\' around the edges\'? Or are you going to make something which stands a chance of being sold to everyone who\'s looking for the \'ultimate\' piano replacement/s.
2. A real piano with imperfections is still an organic instrument with a myriad of
interactions determining how each
\'imperfection\' presents itself at each beat of the music. Play a hard C major chord on beat 1 of bar 1 and you might hear a bit of a string buzz on the bass notes. Play the same hard C major chord exactly the same way on beat 1 of bar 2, and maybe you won\'t hear a buzz at all...
This organic interaction is virtually impossible to emulate in a sampled recording - there are just too many possible interactions to emulate, and no known user interface system exists which would allow the player to work \'with\' the imperfections, as opposed to simply having them turn up unpredictably.
Say you go in and sample the Abbey Road piano, and there is a string buzz (or whatever noticeable and characterful artifact) on, say, C3. You keep it as part of the nature of the piano, and programme it
into the final Giga piano instrument.
Whenever you play that C3 at that velocity, you will hear EXACTLY the same buzz. Not a slightly different one -exactly the same one. When you hear it once, you won\'t notice it much - maybe it\'ll be cool. But play that note a few times within a shortish period of time and the buzz will start screaming \'I\'M A SAMPLE! I\'M A SAMPLE!\' It won\'t be long before the piano annoys the **** out of you and you go back to one which has less obvious artifacts.
I know that theoretically it\'s possible to sample that single C3 many, many times so that you have lots of possibilities every time C3 is hit at that velocity, but as technology stands I think we\'re only in the infancy of truly emulative sampling. It\'s as though sampling up until Gst has been 2D, and now we\'re moving to 3D. There are lots of new possibilities, but the technology requires a hell of a lot of grunt and even more time, focus and money spent by the library developers.
I wonder how much Nick\'s $499 VOTA would have cost if he\'d had to record and programme 100+ times that amount of material in order to include all the possible quirky nuances which might be heard during a live choral performance?
In the end, I suppose we\'re all trying to decide how long the piece of string needs to be, and right now I can\'t keep up with this fabulous flood of meticulously \'correct\' releases, let alone start investing in more adventurous and less correct alternatives.
But you\'re right about what our aspirations should be - a real analog emulation (organically interactive), and actually, there\'s at least one developer who is swimming against the tide by trying to do exactly that (Franz - you know who you are, you madman!).
You make na excellent point about the repeatability of a \'character flaw\' announcing it\'s nature. BUT...
I\'m really new to GS, and don\'t yet know its limitations, or even what the various layers can be programmed to do, but I have a Voce V3 Hammond module that sounds really authentic on record, because part of its character is generated by random processes-hit C3 once and get a big key click, hit it again and the artifact is about 30% as prominent.
Now if I were to try to emulate that in a sample, I\'d start with a clean sample of the instrument, and then add an isolated (computer-reconstructed) artifact layer. (Clean sample minus dirty sample=noise artifact.) Now add that layer back in at a random volume (S/H at each keystrike) and you\'ve got \'character\'
Or I\'m talking out my, er-that\'s through my hat!
But I think it can be done with the tools we currently have, it\'s just a heck of a lot of work!
There was a LOT of discussion on the wishlist for GST V3 which centered around switching between samples in \'horizontal\' layers, which I think could provide a solution to the \'I\'m a sample\' announcements created by repetition of the same sample in a particular velocity level.
Right now most samplers concentrate on switching samples \'vertically\' - ie, from soft to loud samples, or dull to bright, or sustained to pizzicato etc. Sampler interfaces are designed to give the developer quite a few choices on how he can move from a soft sample to the next loudest.
Horizontal switching would switch between samples which are virtually identical within the same velocity level of the same articulation layer. There are a few different opinions about how this should be done - whether it should simply be random, user programmed etc.
Imagine that each velocity layer isn\'t simply holding a single stereo sample, but a pool of 4 or 5 (who knows how many?) samples which all sound \'the same\'. They aren\'t the same though, they are each different samples of the player/s trying to do the exact same articulation. Instead of your C3 FF Pizz layer having one stereo sample, it has four FF Pizz samples, all recorded at the same seesion, all equally capable of standing alone as a good example of C3 FF Pizz. Each successive strike of C3 at 127 invokes one of these four samples.
Again, I think the requests we\'re seeing for this type of sampling are a result of a combination of Nemesys excellent use of HD technology, their sample engine innovations and maturing PC technology.
Even just five or six years ago. No one (apart from Hans) would have seriously asked for samplers which supported multiple \'identical\' samples in the same layer . Most of us were too busy trying to fit more samples into our $1000 dollars worth of 16mb ram by editing our stereo samples down to to mono!
Obviously, the goal posts move as technology gives us more options at an affordable price, and I think that\'s just bloody fantastic.
I started my sampling history with Gigastudio less than two years ago...but I do ALOT of work with small (250k-500k) sample banks...and man am I spoiled by GS. Thanks to those uys for pushig an envelope.
In my constant search for things to say that reveal my ignorance about sampling, I had a thought yesterday that might be worth pursuing as a way to demonstrate just how far I\'ve drifted from rational thought:
If overtones are what we miss, and they are unavailable because of memory limitations, would it be possible to gain the needed memory by doing extremely narrow samples: by creating individual sample sets for each key? Say we took G major? All the notes on the keyboard could be sampled, but only the notes in the G major scale (and the flatted 3rd and 5th?) would trigger overtones. Would there then be enough memory left to trigger a completely different sample set when two or three or whatever notes are struck at the same time, so we could sample chords instead of individual notes?
I know the idea of creating individual samples for each key sounds very limiting, but that limit would seem to be one we would notice most if playing live or playing around at night, and not wanting to switch samples between every song. Since giga instruments are instead usually used for studio work, however, might it be worthwhile to take the 30 seconds to switch samples between songs if the result was a good piano with overtones?
The larger limitation would of course be over the passing tones or enharmonic tones of much contemporary music. Single scale sampling would not be of much use for playing much modern\\contemporary jazz or classical music. On the other hand, might this work for the many other different types of music?
And try Film nior with big band, and strings and other orchestral elements, not to mention electronics. I\'ve got a baritone sustain down to a near single cycle in some patches. It sounds horrid, but its something