Have You Ever Sampled A Real Piano Yourself

[skyyder]

Ok so it's much easier to purchase sample CDs but has anyone ever sampled a real piano on their synth.

 

How is it done, what are the important steps to consider? Do you just plug in a mike drop it inside the head board and record each note . Hey I don't think so that'd be too easy and we would have solved the problem of which synth has the best piano patch by now

 

I think it must be a pretty cool thing to sample a real piano, but then maybe I am wrong cause I have never done it and hence this post.

[Yoozer]

Originally posted by skyyder Ok so it's much easier to purchase sample CDs but has anyone ever sampled a real piano on their synth.

I haven't, and I wouldn't .

 

How is it done

You need a completely 'dead' room. Reverb can always be added later.

 

what are the important steps to consider?

Mainly consistency - besides the obvious volume chance when you strike a note, there's also the timbral change (hence, triple-strike piano's). You need the force to strike the keys to be consistent so you can apply a decent (non-linear?) scale afterwards. You need to sample at the best possible quality - scaling back down afterwards is still a possibility. Get the best DAC's you can afford (rather 2 decent ones than 8 standard ones). Study the physical model of the piano - pedals make a difference, there's sympathetic resonance (which you might model too, but that's a rather daunting task).

 

Do you just plug in a mike drop it inside the head board and record each note . Hey I don't think so that'd be too easy and we would have solved the problem of which synth has the best piano patch by now

Naturally. There's a 'left' and 'right' inside a piano, and a single mic can't capture that properly, I think (has to do with the characteristic). So, you need multiple mics.

 

I think it must be a pretty cool thing to sample a real piano, but then maybe I am wrong cause I have never done it and hence this post.

Take the most logical, scientific approach; leave the whole "feeling" to those who play your sample on a weighted board without seeing what it is.

 

Personally I think there's enough to research in the physical modelling department. It'll save a lot of work with sampling; on the other hand, you'll have an absurd amount of parameters (size of the resonant chamber, chamber material) to adjust, and most likely, it'll eat your CPU alive.

[cheptronics]

I sampled a 1847 Erard grand piano that belongs to a friend. I recorded some notes in stereo, at three different velocities. That was the first sample I made, besides a complete drums.

 

 

There's a big problem in sampling a real piano. If you want something realistic, you have to sample at least two different velocities, in stereo. The two layers are better being crossfaded. Which means that one single note takes 4 voices of polyphony ...

 

I have great piano samples, they sound far better than a Yamaha P-250 for example, but the musical result is often not so good because my sampler (old A-3000) has 64 voices of polyphony, so 16 "actual" voices with a two layer stereo piano sample bank ...

 

I saw lots of 5-layer or even 8-layer piano samples, but I don't think anyone use them in practise because the most performant computer samplers, like Kontakt, have at best "256 stereo voices", so an actual polyphony of 512, which is weak for such samples. Am I wrong ? Maybe the crossfading between layers is more optimized in Kontakt than in the A-3000 ...

 

Also, it's difficult to emulate sympathetic resonances unless you are using a software like Pure Data.

[Synthoid]

I agree....it's too much work and the storage space for those samples is huge.

[studentOfSoong]

I use a 6 layer piano in Kontakt, a 12 layer rhodes and a 16 layer wurlitzer. No problems with polyphony, with the piano I can play 40 notes or so, before the note-eating algorithm needs to help and about 64 with the (mono) rhodes and wurlitzer samples. No problems.

 

However, sampling a piano so that you get sounds that are actually workable in live situations is more difficult than you would think. The sounds in Yamaha P series are adjusted to work with standard PA's and keyboard amps, while playing unprocessed piano samples through those things end up unplayable in my experience. You will need really good amplification, and possibly some nice (tube) preamp to get unprocessed samples to sound well.

 

And indeed, you'll need good mic's and preamps, an anti-accoustic noice-free recording room, a lot of recording experience, a lot of time on your hands, patience etc.

[Dharmabum]

Grand pianos are stunningly complex instruments. There's plenty of digital pianos with pretty nice emulations, that will pass for the real thing in a lot of situations. There is still no replacement, however, for the feeling of the strings resonating back through the wood into your fingertips; the sound waves of the strings themselves eminating from a broad surface area compared to just being projected out of a couple speaker cones.

[skyyder]

Phew, I didn't know so much goes into sampling a real piano I guess we need to sometimes appreciate what we've got in terms of whats available and the technology that goes into making those samples we purchase and the waveforms on our synth. Not that they shouldn't try and make them better, but it seems like a whole lot goes into providing us with those piano sounds we clamour about each day:D

[Dreamer]

I have never sampled a real piano (I mean the acoustic instrument), because I don't have access to one and -even if I had- there would be huge problems with the quality of the microphones, their correct placement, etc

I have, however, sampled pianos found on sample CDs and the results were good; the best piano I have found so far (between the medium priced ones) is the one found in the McGill University library: they have sampled the entire 88 notes of a Steinway grand.

I have that to say that I was satisfied with the final result, because I was able to obtain good loops, but when I bought my first digital piano (a Yamaha PF P 100) I felt a little ashamed; so I decided to sample that one too, but the final result was not on par with the original samples, maybe because (due to lack of memory) I had to sample just one velocity level. The original samples had obvious loops, but I was unable to duplicate them because Yamaha applied envelopes to the original samples after looping them, so -due to the volume decay- the final point of the loop had a much lower level compared with the starting one.

All in all, I would say that it's possible to obtain a good sample from a CD or a digital piano, while I would disadvice anyone to sample an acoustic piano (unless you work in a recording studio), due to the microphones problems I mentioned before.

BTW, I own an Akai S 2800 sampler.

Regards

[Dreamer]

Just forgot to add: another BIG problem when sampling an acoustic piano is the difficulty of playing all the notes at exactly the same velocity; with a digital piano all you have to do is to send a midi note from a sequencer at the desired velocity.

[cheptronics]

 

Originally posted by studentOfSoong I use a 6 layer piano in Kontakt, a 12 layer rhodes and a 16 layer wurlitzer. No problems with polyphony, with the piano I can play 40 notes or so, before the note-eating algorithm needs to help and about 64 with the (mono) rhodes and wurlitzer samples. No problems.

 

That's great, because on the A-3000 it's only impossible to do this. I have lots of work now, but then I will program a minimal polyphonic sampler under Pure Data. Kontakt looks good (I'm trying the demo currently), but I don't know if I need to buy this. 40 notes is great, of course, but you can play 128 notes on a P-250, for example, because it's particularly optimized for piano sounds while Kontakt is a generic sampler.

[studentOfSoong]

Originally posted by cheptronics That's great, because on the A-3000 it's only impossible to do this. I have lots of work now, but then I will program a minimal polyphonic sampler under Pure Data. Kontakt looks good (I'm trying the demo currently), but I don't know if I need to buy this. 40 notes is great, of course, but you can play 128 notes on a P-250, for example, because it's particularly optimized for piano sounds while Kontakt is a generic sampler.

 

The P250 has smaller samples, which it can play directly from RAM, while my 4.5gb sized samples don't fit in my laptops RAM, so Kontakt needs to use disk-streaming to play the samples. The disk-streaming is the limiting factor here. With some sample processing it should be possible to get the sample smaller and thereby increase the polyphony. However, the 40 polyphony doesn't get me into problems, usually.

[the_resonator]

i have a baldwin grand at my old house that i have sampled on numerous occasions.

I never did so with the intention of ending up with a realistic sample.

Most of the times i did so were more than 8-10 years ago, and sampling anything was fun and "experimental" (feeling old now..)-so they were samples made to be used for various purposes, not to masquerade as a real piano.

 

To get a decent and usable sample bank of a real piano- you are tlaking about finding a good dead room, getting the piano in there, mic'ing it with the right mics in the right places (probably more than one mic and more than one type..condensor, standard cardioid, boundary, etc.). You'll need a bad ass mic pre, a bad ass equalizer, etc.. You'll also need to sample at a bunch of different velocities, as well as each key with pedals up and down because the manipulation of the sustain pedals changes the character of the sound. I mean, it just goes on and on...

 

I think, now that people can actually make a living making samples and libraries, you should let their years of experience and studios full of dedicated equipment take the lead.

 

or not.

 

Now that i think about it, you might want the sample to sound like you made it rather than a professional.

 

never mind.