Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Recorded Audio

Why, you ask?I didn't like the sound at all. And it's been revealed as a digital sample.
So @pro is right about one thing. The sampling was faithful. The digitized piano sounds crappy, just like the original.
But he's wrong about one thing: That's not a contradiction.

@propianist: you said you used the same hw to record the 2 different instruments. But it was the same location too? The same recording studio?
Apart from the more precise and deeper bass notes (of course, it's a Steinway D!), to me the Piano n.2 sounds more "dynamic", "bright" and "open", but if I listen more carefully I also think the pianist used more pedal on the upright, compared to the grand version, especially in the central part of that excerpt.

P.S.: I think MacMacMac meant that the piano n.2 sounds very dry. Usually you imagine a Steinway D in a big hall, so you expect some nice natural reverb. But if you recorded it in a recording studio full of soundproof walls then that type of dry sound is to be expected, like it was generated by an (high-quality) VST with reverb disabled.

Ok, but I was asking WHY do you think it's crappy. What sound characteristics do you hate/dislike in that upright piano sound (apart from the simple fact it's an upright). Maybe are you saying that you dislike uprights sound timbres in general, or that you dislike that particular upright sound? And if this is the case, what upright piano do you think is very good sounding? What characteristics it should have to sound good in your mind? Just curiosity....

Of course, if you would like an upright to sound like a grand, then there would be an issue at the base of the reasoning...

Good to get the answer, this was a fun ear exercise. What surprised me maybe most is that a 100MB piano VST was mistaken for an acoustic by some of those who took a guess.

I would have understood if it had been a high-end VSTi, something like the VSL Boesendorfer, but not a freebie soundfont or whatever it was. I'm curious to know which instrument was used in the first recording. Was it a known piano VST?

MagicPiano
No, the two pianos were not in the same building or location, as I said.

TheodorN
The 64MB sample was a Kemble upright piano (Kemble have been owned by Yamaha since the 1980's) recorded in a domestic living room with carpet and soft furnishings and was multi-sampled and edited by me when I was a much younger chap, approx 22 years ago.
At the time I though it sounded more realistic / lifelike than many of the vanilla commercial digital piano modules and stage pianos of the time.
So thinking the results were quite promising for a small audiophile 64MB library, I decided to go into a local studio to record a Steinway D - my dream piano. The studio room was much larger with wood tiled floor and high ceilings. The result of that sampling session unfortunately was less impressive to me, perhaps because the Steinway had much, much longer sustains with evolving timbre so really couldn't be satisfactorily captured in just 64MB (to fit my Yamaha EX5R rack sampler of the time.) So after that, I kinda gave up trying to multi-sample pianos myself. Nowadays there's so many fairly good commercial VSTs that it's quicker and easier just to keep shopping around buying those VSTs from time to time hoping to find the holy grail one day, whereas sampling your own piano takes forever to manually edit all the notes in the DAW. I furthermore think that because most commercial companies these days use automated software scripts to batch process all the note samples that certain bad notes fall through the cracks undetected. My old fashioned approach was to listen carefully to every single note individually and edit each one by hand the hard way, but that yields much better results.

MacMacMac
Okay, so you do NOT see the sheer contradiction in your statement. Good to know.
I'm sorry you think my upright piano sounds crappy. Everyone has to start somewhere and that's all I had when I was a teenager. If I could afford a Steinway I would buy one.
The Gateway Studios Steinway D recording session cost me £50 an hour for 4 hours and in such a limited short time, I just couldn't get everything recorded how I wanted, in terms of getting enough different velocities captured for each key and choosing best mic positions, etc. That's probably why that Steinway D multi-sample never impressed me as much unfortunately, and I probably wasted my £200 there, but it was an interesting learning experience. I think I'd need many long days in a studio to get an excellent multi-sample of a Steinway. It's not an easy job. It's not a cheap job either when you're a teenager.
At least that's one advantage of recording your own domestic upright piano in your own home - you've got ample time spare, no studio clock ticking, but the 64MB sample RAM restriction did obviously render the end result a bit "crappy" it's fair to say. I believe the actual instrument sounds and plays much nicer in real life.

To me, they both have that out of tune ‘saloon piano’ sound

Thank you for these explanations, propianist, an interesting story. Just shows that sampling a piano is no easy task, takes a lot of tedious work, and precision.

Hi folks! My first post here, and my two cents.

1. What are the things that give it away?
You know, I usually can tell it's an acoustic due to a simple but striking feature of APs: 99% of time, they are out of tune. A homemade (or semipro for that matter) recording that does not have at least one clash...well, it was made with a DP, no doubt about that.

(Even in the realm of professional recordings...you listen strange things happening: the grand piano can be sometimes an unpredictable beast.)

Another indicator is pedalling. Although I've seen this particular point improving quite a lot in the last years, it is what generate the most noticeable artifacts in digital processing of a piano sound.

2. About using real MICs, etc. this will only decrease the quality of what you already have on the package. You'd have to have a stellar studio equipment (and a stellar studio room) to make any resampling/remicing/reanything justify the effort. And I doubt it will improve things the slightest.

3. Would it matter to me?
I'm an enthusiast of DPs since the early Clavinovas of the mid 2000s. I think we are approaching the 100% asymptote of emulation at a fast pace, and we are very, very near to the point of being indistinguishable for all practical purposes. Even before that, I'm about music, not fetish: I prefer to listen to someone playing very well with a Casio keyboard than to listen to this mainstream classical garbage played with the most unique Steinway, at the most unbeliveable stage, with the most cutting-edge audio production...oh, I'm annoyed just to write this...

My very best regards,
J.

Oooooh! I feel you, J. I bet you have someone in mind, when you typed this... Who's it? c'mon now.. don't be shy... Share it! Share it! Share it!

That's a major advantage of digital pianos: the strings of a note can be correctly tuned to identical pitch.
Makes no sense to reintroduce that flaw of acoustics' incorrect tuning.

The keyboard at home has speakers on both sides. Audio recordings need to be pointing towards 1 of the speakers for optimal pickup. Otherwise would get a lot of recording noise. The sound of an acoustic is more evenly distributed in the room so don't notice black spots in recording pickup.

Besides using a microphone, haven't been experimenting with connecting an audio cable from the keyboard headphone jack directly to the computer. This is supposed to eliminate background noise in the room.

It's an older keyboard with a reasonable piano sound. The reverb is not quite the same. The long notes don't hold as well without using a bit of sustain pedal.

Have been using digital for a while and only record on acoustic at a local conservatory when available. During music lessons the teacher uses a Yamaha Clavinova. She doesn't think learning on a DP is an issue.

Dear Josh,
Actually, I was thinking about an unfortunately large list of pianists: the commercial realm of hodiern art is a giant recycle bin, and since it's very difficult to sell old trash - mainly because it stinks of mold - people resort to progressively more desperate means (which includes all this fetishism of super-ultra-hyper cool stuff that does not add a penny to the music being made).

However, it is understandable: how do you promote something to sell if you have free versions available that offer the same experience and are subjectively better? Marketing 101, you overhype what is not available to everyone. And it's surprisingly effective! Apple keeps making billions with this beaten formula...

Now, coming back to the subject (sorry I digress a lot), I think we are freakingly near to the point where a piano album made at home with little to zero budget (other than the gear itself, but I mean production costs) will be sonically identical to anything a triple A label can offer, other than having a beautiful face on the cover and a ton of formidable stuff trying to justify the sales.

And to someone who thinks that it's not the case, when was the last time you bought a CD and when was the last time you digged into an unknown pianist YouTube channel? And what experience did you enjoy the most? Hip hip hurray to DPs, VSTs and DAWs: here lives the future of art music.

Dear Burkey,
I think here lies the major challenge to making a digital piano sound 99.99...% authentic: the shape of the sympathetic resonances of a grand piano. There is an acoustic law at stake here: if you get three identically tuned sources, they don't add the same as three very very very very slightly "out of tune" ones. The reason, as modern piano wrote, are the beats (not the beets, it's important to make it clear...). And here is the part of approximating an infinite sum to save computational resources makes a difference: we musicians - and the well-acquainted lay persons - are like dogs, listening to subtleties that are sometimes difficult to measure.

However, I think we are closing the gap: combined approaches of sampling and modelling, and shear horse power, are the key IMHO.

As a side note, I agree 100%: having a well-tuned piano is the major advantage of a DP, because playing out of tune - as is the usual case with acoustics - is a nightmare and the reason I sold my upright many many years ago...

Best regards!

If a realism was the goal, then we're not closing the gap at all just because of one area alone: Sympathetic Resonance. Going by Wikipedia's definition here:"a harmonic phenomenon wherein a formerly passive string or vibratory body responds to external vibrations to which it has a harmonic likeness." I haven't found a DP or VST that could implement this live. This is much harder to replicate than half-pedaling, IMHO.

If an ideal piano sound is desired, doing away with the hammer strikes , pedal sounds, etc., then yes we have been closing the gap...

Well, that's not correct: inside the eprom of your typical digital piano engine, you don't have a different sample for each string of the original acoustic instrument... You have no more than 88 samples * n. velocity layers (but there could be other samples for the release and various resonances). So, in your typical digital piano engine (unless it's a Roland V-Piano) you cannot tune separately the 3 strings of each note. This means that if the original instrument had some not well tuned note because 1 or 2 of the 3 strings of that note (in the original acoustic piano) were not well tuned, you'll never be able to fix the issue with the options inside your DP. Either you like how the 3 strings of each key were tuned in the original instrument, or you have to buy another DP... (not many DP manufacturers allows you to put new samples inside the eprom!)What you call "a flaw" is just one of the many things that make an acoustic grand piano like a Steinway Model D or a Kawai SK-EX sound so beautiful and natural.
Do you know what happens if you take out from a piano sound some natural effects like the "beating", the "dispersion" (inharmonicity of upper partials in a vibrating metallic string), the "double-decay" and many others you can read in most research papers about piano sound modeling? Probably something like this:

BTW, the Roland Juno-60 was a very cool synth...

Modern digital pianos have all kinds of resonances where "formerly passive strings" resonate: damper resonance, string resonance, undamped strings resonance...

...and as a bonus even cabinet resonance.

@cloth-meister: Aren't all of those resonances just synthetic substitutes?

...but still no Sympathetic Resonance.

I gave a link of the full definition of Sympathetic Resonance in my original post here: Post#3117575.
Where'd you get yours? Can you name one DP or VST that explicitly says they have Sympathetic Resonance? +link of course

Where's the link?