Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Why do samples sound different from an acoustic piano?

I am in need of some basic advice please.
I understand that reproducing anything exactly is nigh on impossible and that the only way to get the exact sound of an acoustic is to play an acoustic. But what I am wondering is.. if I use say Vintage D ( which I have) or American D (which I thinking of buying) , is the end result mostly dependent on my PC/sound card/headphones or on the samples themselves? I would love to be able to reproduce the lovely ringing tones of an acoustic Steinway but whether in my setup or listening to demos I get the impression that something is missing - maybe higher harmonics or as the result of compression? It just sounds flat.
I have a PC system greatly in excess of minimum requirement specified but which uses Intel onboard sound. I am able to set any ASIO4ALL buffer size/number I like without any problem with dropout or latency.
Would an eternal sound card help at all? I am loathe to spend a lot of money on a new sound card or headphones if it will not help.
I am using a FP-7F as controller and Roland headphones (£50 only)

I'd be grateful for any advice you may be able to offer. Thanks.

The closest we are able to get currently to reproducing the sound of an acoustic piano is via the use of actual recordings of a piano (sampled) or by using a mathematical model of what happens when a felt hammer strikes a string inside of a wooden box (modeling).

Two of the most popular products of late are Ravenscroft 275 (a multi-sampled piano and Pianoteq 5 (a modeled piano).

It is generally felt that making many and large recordings of a piano at various volume levels key by key produces a sound that is more similar to a real piano. However, mathematical modeling is thought to do a better job or producing variations in sound that naturally occur in an acoustic instrument.

I would suggest listening to demos of both kinds and where possible installing a demo or trial version to play and hear for yourself. In many cases and situations you would be hard pressed to tell the difference between a recording of a sampled piano and a recording of a piano. However, in solo play - where we are able to listen for tell tale signs of mimickry - a discerning listener of course can point out the differences.

The software solutions are currently offering better specs and reproduction than most of the digital pianos on the market (although not all). And yes... your listening environment and sound reproduction equipment will play a large part in how "good" things sound.

At minimum a proper pair of headphones is required. Even when using the on board audio card in your PC. Here is a guide on how to select the appropriate headphones for your listening pleasure. Full range of sound reproduction within human hearing is critical (look for phones that reproduce at least 20htz-20khtz, are comfortable to wear for long periods of practice, and that YOU think sound good - so try them if you can).

A better sound card than the stock chipset on your motherboard may improve sound quality for a discerning listener because one may have better digital to analog conversion and a better headphone amplifier. Plus, it may be able to function at a higher sample rate, at least matching the sample rate that the sampled piano was recorded at so there is no sample rate conversion happening. Ravenscroft 275 when installed in full is 35gb uncompressed. The download version is 6gb because it compressed in lossless FLAC format. FLAC is not a lossy compression format, so the only difference between the two would be CPU processing requirement.

Then, if you tire of wearing headphones you would consider a proper pair of powered monitors which should be selected with just as much care and attention to detail. They should be the right size for your room, be mounted at the right height and away from the wall, and separated from what they are mounted on using isolation pads.

How deep you get into this depends upon how picky you are. And there are plenty of picky people here, so no doubt others will have opinions to share.

Sound is indeed quite dependent on the audio chain (digital to analog converter, headphone amp, the headphones themselves). The PC itself, not so much if basic specs are right. We don't know what quality soundcard and which headphones you are using, thus it is difficult to be sure. A high quality external soundcard/headphone amp, and high quality headphones may indeed improve the quality.

Of course the whole setup (from the keys to the headphones) contains a lot of variables which may contribute to subjective frustration. Latency, or inadequate key actions and/or touch curves, may be culprits as well. Tell us a little more about your setup.

And yes, if you want the full impression of an acoustic, play an acoustic.

I'm not sure I understand this point. FLAC compressed samples will sound identical to uncompressed samples. It's not necessary to A/B the two because the data - once played by the controller and output through the computer - is the same.

James
x

My apologies, I was covering a lot of questions. FLAC is not a lossy compression format. Only difference would be CPU requirements to play back. Edited for accuracy and clarity.

Thanks for the clarification Elmer!

- off the top of my head, two reasons,

* no one's got the sampling quite right yet. It's close but there's always something overlooked and, even with feedback from users and beta-testers, I fear the developers run out of steam and do the minimal repair work and debugging rather than instigate an overhaul of their first effort....which is often what's required.

* the sound stage from monitors and headphones - even the best of their kind - isn't the same as would be experienced from the expansive, resonating wooden soundboard found in acoustic pianos.

I think the first thing I'd upgrade (and what might make the most difference) would be the Roland headphones. There are many "What's the best headphone?" threads here already. You'll probably spend more than 50 pounds on a new pair.

The second thing would be the computer's internal sound card. External "audio interfaces" -- e.g. the Focusrite Scarlett 2i2 or 2i4 -- are reputed to give audibly better results. [I'm still using my internal sound card -- just stubborn, I guess.]

There's lots of discussion of "Is an external interface better than the built-in soundcard?" on the Pianoteq user forum, which is accessible to everyone:

http://www.forum-pianoteq.com/viewforum.php?id=1

. Charles

PS -- I use Shure SE215 earbuds, and Sennheiser HD-280 headphones. The Sennheiser HD-380 phones are reported to be better than the HD-280. Choice of headphones is very subjective.

I would say that once you have eliminated latency as an issue, I think the benefit of a different interface is questionable, and the benefit of a different computer is non-existant. OTOH, different headphones (like different speakers) can make a huge difference.

Though on a side notes, has anyone ever gotten a really authentic vibe through headphones? I wonder if they can ever match what you can get through speakers, even if only because of the way you can get real room reflections with speakers (as you would with an acoustic).

I'm not sure, Scott. I've been hunting for a good audio interface for a long time, and been through a lot. They all sound different, and for sure there were some I preferred over others. AD conversion, sample rate and quality of headphone amp as well as what's being delivered at the monitor outs can be very noticeable. When I'm working on a mix/master these days I use several headphones as well as speaker types for comparison - from the cr@ppiest to the best I can get my hands on to make sure it all translates well. At some point your comparing minute differences, but not in the basics. I've never got desirable audio quality from a motherboard's stock audio interface, worse on laptops than desktops generally speaking.

I'm always amazed when people suggest using headphones as a primary option. The only way to get a realistic sound of a piano (or orchestra for that matter) using headphones requires binaural recording. Binaural recording is the process of embedding microphones in the ear canals of a dummy head and placing that dummy head in the acoustical environment (a hall, theater, living room, ... whatever) that you wish emulate. In practice its more complicated than it sounds because the microphone characteristics have to be designed to capture a flat response in an unusual acoustic "ear canal chamber" and the headphones must also provide a flat response to the ears (which is not the goal of many headphones). But research over decades has shown proper binaural recording to produce superb results when using the proper techniques and equipment.

Binaural recordings generally sound awful when played back over high quality speaker systems because the recordings contain the acoustical properties of the environment they were recorded within in addition to the frequency, phase and other physical effects of the "head". When they are played back in a room over speakers those properties are convolved with the acoustic response of the room as well as the speakers frequency response and directionality properties. i.e. they don't sound right. It's rather like making a recording in a "very live" room and then during playback adding convolution reverb that models a completely different room.

Bottom line, the acoustical characteristics of stage pianos and such, are normally designed to sound best in larger environments over external speaker systems (where flat frequency response is not their primary goal), and certainly not headphones. Pianos for home use are normally designed to sound best using their built-in speaker systems in smaller rooms that would typically be found in homes. Only a few products have been produced with additional, separate binaural recordings (sample sets) actually designed to optimize headphone use. Hence, when someone "reviews" or comments on how a software piano (or DP) sounds using headphones I just cringe and ignore it. I never use headphones when playing my pianos unless it's the dead of night with people sleeping in the house and I just can't stand to wait for tomorrow.

On interfaces, I don't claim to have superhuman hearing nor am I particularly wanting (maybe I'm missing half an octave at the top end through creeping decrepitude) but I can't discern a dot of difference between either of my audio interfaces (UR22 and the dirt-cheap Behringer - UAC222 or something like that) or the internal sound card.

With headphones, I usually add a dash of small room reverb - mostly early reflections. It needs to be subtle but I find it so convincing I could swear I'm listening on my monitors and have to remove the headphones to be sure the monitors aren't making a contribution.

That really intrigues me, Dire Tonic. What kind of speakers are you using and are they setup in an acoustically friendly environment?
And what kind of headphones are you using?
I have a pair of Koss Pro4AA Titanium Professional Stereophones (very impressive name, surely). I have no idea of how "good" they are considered to be, but surely some distance up from the bottom of the heap.
I've used my monitors exclusively for everything after I bought them. It takes quite some time for the ears and senses to "tune" into them and learn how to use them,and what is possible to listen for. They sharpen your listening senses (something I didn't understand before getting them).
And now finally to the point.
One day I thought I would use the headphones for checking out the details of some material I had put together, to see if it would show up better in the headphones. And it was quite a shock to hear how bad they were! I found them totally useless and got them off pretty quick!
I still use them while recording vocals and they work well for that, because then I just use them as a guide and reference in timing and place and I'm not really listening to sound quality.

Have you tried using your FP7 piano with the software piano together? You`ll get the effect of multiple strings at work here. I do this in my own set up and the results are impressively acousticky . . like a new dimension. I use Pianoteq 5 (K2 under lid) through the DP speakers. Deep and powerful.

I'm using Focal Alpha 50s, carelessly in a non-treated room. They're nearfield and have no offending resonances I can hear. I like them a lot but they're way down the league compared to your Genelecs.

I use sennheiser HD595 and, for comfort, an old cheapo pair of PX100s. I'm not aiming here to mimic the monitors although there's a sense in which I think they're exceeded. I participated in a couple of ABF recitals and was struck by the AP entries. Most were recorded willy nilly but they all had that AP sound of which, to my ears, the roominess forms a vital part. I'm trying to get that 'informal' recital sound - the big hall is faintly ridiculous for the kind of stuff I play!

The PX100s have helped add to the illusion since they weigh almost nothing and are surprisingly uncoloured. I'm still experimenting with the room effect but I usually prefer something bigger than the room itself while minimising the reverb in favour of reflections. I use vst plugins inside a host rather than the onboard effects that come with the libraries.

I'm in two minds about layered pianos. There's a tendency here and there to get phase cancellation or, worse, with a tuning anomaly, a chorusing effect reminding me of double tracked pianos on pop records from the 70s. On those occasions when I'm bugged by the shortcomings of a library (often!) I'll try a layer. After a while that bugs me to. At that point I usually watch the television.

I messed around with layering pianoteq and a sample player.
For VERY SLOW playing there appeared to be a "richness" of sound, but fast staccato single notes seemed to have an echo.
I didn't get into it, but I think the delay of the sample player is significantly greater than pianoteq.
I can get a similar effect with reverb and delay lines, so I doubt the utility of this layering - other than that it MAY sound a bit like two DIFFERENT pianos and two pianists with one playing a bit later than the other.
It probably becomes annoying after a while, though its judicious use in post processing recording might be useful.

Let us not forget that even a top tier LARGE acoustic piano in perfect tune can sound terrible in SOME settings.
Hard furnishings, between hard parallel walls, low hard ceilings over hardwood floors, etc.
It can just be TOO MUCH piano for TOO LITTLE space.

How close to a physical model D modeling or sampling could come is irrelevant in THIS house.
A REAL model D would make the walls ring for an hour after closing the lid.

I too noticed this. I haven't tried it with pianoteq but there was often noticeable albeit close flaming when layering some of the earlier libraries. Piano in Blue which has a lovely sound but speaks late was a serious offender. I've found the latest crop all strike well and together. I'm sure these days the developers are acutely sensitive to low-latency demands.

In re-reading the root post - and then the replies...
I get the idea that we may have been "talking past" the original poster.

My "GUESS" is that what is missing from the desired "lovely ringing tones of an acoustic Steinway" that the O/P wants is the room (hall).
i.e. the sample libraries have been recorded "dry" and some ambient environment needs to be added to bring those "flat" samples to life.

I was about to say "dehydrated samples" and maybe that is a fair analogy after all