I sampled a Steinway grand today !

[mickrich]

I sampled the Steinway grand piano in Windmill Lane studio 1 with my class today. We recorded 4 notes per octave at 3 velocity layers with a pair of Neumann 184s.
Very beautiful natural sound. I can see myself using this a lot!
Here is MIDI file of Brahms running through it.
https://dl.dropbox.com/u/2133088/Brahms_WIndmill_Steinway.mp3

 

[lanky noob]

That piece sounds total final fantasy to me

samples sound awesome though dude! where abouts were the 184's positioned? both close mics or one close/one room?

(In case you hadn't guessed I have no fucking idea about recording pianos)

Edit: Just realised who you are, thanks for the log style multi and something with the file name "end trax 165 bpm" you put a while ago, helped me learn and improve with my mixing a hell of alot /end asskissing

 

[mickrich]

Mics were set up about 2 foot above the piano as a spaced pair with the lid off the piano.
Glad I could help with the multis

 

[Ericlingus]

how were you able to have it play this song with only 4 notes per octave?

 

[Mutant]

how were you able to have it play this song with only 4 notes per octave?

All samplers can do that.

You can tell the sampler which wav it should play on which note and stretch the sample region to neighboring notes and play it slower or faster there to make a perfect semitone change on every next key.

 

[amarshism]

Bad ass

 

[TheWinterSnow]

quick question, how much more processing on the software side of things needs to be done to get more samples per octave and velocity layers? Was the reason for such few samples due to time constraints?

I think if you went mic crazy and sampled every note of the piano (12 notes per octave) with around 12-16 velocity layers per note you could have a product on your hands that could make a lot of money, if for say, you used kontact, it could provide pretty successful.

My one complaint with the sample sound is that you can tell that the mics are far away, there just isn't enough low end and it lacks an in your face kind of sound. Still sounds great.

 

[mickrich]

4 notes per octave was due to time constraints.
Not any harder to sample every note, just takes longer.
I would need a pro piano player to get 12-14 velocity layers.
The mics were actually close enough, the Brahms piece is very low velocity.
Here is John Lennon's Imagine.
https://dl.dropbox.com/u/2133088/Imagine_WMS.mp3

I agree that sampling this piano could be worth money.
I will have a chat with the director of Windmill about setting up a partnership to make this. Conversion to Kontakt is pretty easy but I would need to get a graphic designer to do the snazzy pics you get with kontakt instruments.
I am actually amazed at how good this sounds for the 2 hours we spent doing it. I also used the mic pres on an mbox due to being in the live room with the class. If I was doing it for real I would use the Neve desk.

 

[Dan R]

I was actually about to post that I would pay good money for a sample set like this. It sounds great!

cheers,

 

[Skaldir]

Sounds good, a bit mechanic.
You really need to use omnis though with a piano. Thank me later
I think the real challenge at sampling a piano is how you get that sustain pedal working. That's still the point where you can tell a real from a sampled piano. There is happening so much when you press the sustain pedal, and to my knowledge nobody yet could really recreate it.
Awesome class project though

 

[-Noodles-]

Sounds good, a bit mechanic.

I'm agreeing with this. The samples do sound good though!
If you could get a bigger version of the project off the ground there are always people looking for more piano samples.

Thanks for sharing.

 

[Fifth_Horseman]

That must have been a cool experience! The tracks sounds pretty nice, especially considering the little effort involved here.

But before you start investing a lot of time and energy into this project in hope of commercial success, remember there is ample competition in the field of piano sample libraries - especially of Steinway D concert grands, which are probably the most popular piano and thus most sampled.
Just to name a few bigger names of Steinway piano sample libraries: there is Galaxy II Vintage D (for Kontakt), Synthogy Ivory II German D and American D (with an own standalone player and VST plugin), Garritan Steinway, East West Quantum Leap Pianos etc. There are also plugins that use modelling instead of real samples but still imitate real pianos (i.e. PianoTeq D4 for a Steinway sound).

As Skaldir mentioned, the big challenge is not the sampling of the notes themselves, which is basically a diligent but routine piece of work except for the proper setting up of the mics. It is the implementation of all the string and cross resonance that builds up when using the sustain pedal. Different makers of piano libraries utilize different methods to achieve this, like separate resonance samples or a modelling approach combined with samples.

Just to give you an idea: The newest library from Synthogy Ivory II (the "American D", sampled from a famous 1951 New York Steinway concert grand) has all notes sampled at 20 velocity layers. The library takes up 49 GB of space, and that's only for the base samples, as the Ivory engine uses modelling to create the various resonances occuring in a piano. I believe it is common in these sampling sessions to use machines to trigger the keys, to get consistent velocities for all notes on all layers.

So, as far as I can assess your current situation, your product might sound pretty nice overall but would fall short in key areas like resonance implementation and velocity consistency and thus not be able to really compete with the big names.
Obviously, another piano library does not hurt and there will surely be some buyers, but your product would probably not stand out. For that you would need either unique features (like a new, better approach to resonance modelling) or sample a really exotic piano that is not featured by the big names.

So, while I definitely don't want to discourage you from doing it at all, I'd suggest you keep your expectations about the outcome reasonable.

 

[MatrixClaw]

^I'm sure he wouldn't be charging anywhere near the cost of those sample libraries, either. Sure, those are great libraries, but not everyone has $300+ to invest in piano samples.

Much like ZomBass, he could undercut the competition by a significant price, because he's only offering specific samples, as apposed to a huge variety of instruments, and probably do very well from the online communities.

 

[Fifth_Horseman]

Yes, that's another possible niche, offering a superior quality/price ratio.
However not every library costs $300 - that's rather the upper limit for the libraries with multiple pianos, most of the ones I named above are actually significantly cheaper. Several libraries from Native Instruments are priced at around $100, some below that. And there are even free piano VSTi plugins out there like Piano One.
Therefore I stand by my comment that there is ample competition in the market.

This is to be considered if the OP wants to take this from "cool class project with pretty nice results" to commercial product level.
To "do very well" on the online communities I would expect his library would need to offer something unique to stand out, otherwise I'd expect interest to be rather limited. Which means he has to consider beforehand how much effort to put into the project, and also to what target audience it should be aimed.
For example, without any resonance features you can already pretty much count out pianists who want to actively play their software piano or "reamp" their previously recorded MIDI files for solo piano pieces (think online communities like Piano World).
So that basically leaves music production forums - here it is more important how well the samples sit in a mix and how well the sound can be shaped.
Which then brings up questions like: How much ambience should be caught during the sample session?

Again, I don't want to discourage from expanding the project. I'm just not that optimistic that there would be "a lot of money" in doing this, as was hinted at by other posters. The project might still prove worthwhile of course, which again depends on where the OP sets his expectations.
So far he did meet them with the first instance as he sees himself using the created set in the future, so that's a win already.