Noise

[BogdanH]

...
From my experience, there are two musical worlds:
(1) Playing live at home or on a stage
(2) Recording oneself
...

-I completely agree on that (and on the rest).
The difference is not only in how we do it differently, but the audience also perceives the music differently. In live event, nobody will notice slight distortion on some instrument (unless it's hurting the ears) -the only thing that's important is, that the music (as a whole) is good.
However, when we listen the music on good audio equipment at home, we become more analytic and receptive to details.

Bogdan

[Oymmot]

There is no defense of the Yamaha keyboard.
You won't find SEK 50,000 under the bed.
When it also applies to the previous models with the same problem and they continue without improving.
No Yamaha has fallen in my eyes.
Sorry, but there won't be any more Yamaha keyboards for me
Tommy

[ton37]

'If' it has been around since the Tyros, then it surprises me that it has never (as far as I know) been a concern?
It is known that the keyboard has a hefty price, but I don't think that has anything to do with it. In short: a unique finding by a single person, it seems...because it is not further confirmed?? 

[RoyB]

Ton

Not unique to a single person.

As I reported above, I have often experienced 'distortion"-sounding strings, flutes and clarinets when making Tyros5 recordings and had to find work-arounds for it. I always accepted this was an inherent (not so good) characteristic of Yamaha's sounds for these instruments so just got to live with it.

[BogdanH]

Well, it has been mentioned few times before, that strings on Yamaha's aren't really that good... it only comes down to what makes them "not that good". I quite like listening classical music (when I'm in the mood) and in my opinion, strings are far from good. They're usable as background orchestra, but not as leading instruments -just my opinion.
I only have Yamaha and so I can't say if other brands have better strings.

Bogdan

[ton37]

Ton

Not unique to a single person.

...... I always accepted this was an inherent (not so good) characteristic of Yamaha's sounds for these instruments so just got to live with it.

You said it and that's what I wrote. It's not a 'bug' or 'technical faillure', it is how it works .... and one decides whether he/she can live with it or not 

[Oymmot]

Hey all.
These are my views and I know others don't think the same but still want to clarify this about sound.
I want to tell you my views on the sounds that are available to install in from Yamaha as Expansion Pack. To my ears, no sounds can compete with those already in the keyboard.
There are also so-called sampled sounds to buy online and these too are below the originals.
That then the sounds found in the keyboard are better and worse is of course completely clear.
Bought 5 sampled string sounds that in the advertisement on YouTube would be completely revolutionary for SEK 1000. There is so much reverb on these string sounds and it is not possible to influence downwards only upwards and it is already too much. Plus the sounds are not soft but hard just like the newer string sounds found in the new Yamaha models.
For nothing!!!!!!!
My conclusions from this, do not have superstition to what you hear on YouTube and similar places. Everything must be listened to in front of your keyboard.
You can be duped by everything that is available online, but when you get home and listen, it works in the vast majority of cases.
Tommy 😢😭😤😠

[BogdanH]

hi Tommy,

...
Bought 5 sampled string sounds that in the advertisement on YouTube would be completely revolutionary for SEK 1000.
...

There are not many companies that can afford recording samples of an orchestra -because it's extremely expensive. What we get for 1000SEK (~88€) is usually a modified subset of recordings made by some other company. The question is, what modifications have been made. If reverb/ambient/echo is applied on samples (to sound real), then the result is final, because we can't take that away anymore -in short: such samples are useless.

A high quality library can cost 300€ (~3400SEK) or more, having the size of 20+GB. Btw. I have samples of Yamaha CFX piano which has totals size of more than 120GB -yes, piano only! Anyway, such samples are not modified sound wise and in this case we can apply any modifications later on keyboard (in voice editor), to make it sound to our liking.

Every instrument is specific and many times the problem is not the quality of the samples.. the problem is usually the sound of samples. And when we search for strings, we usually look after specific sound (mellow, aggressive, gentle, violins only, violins+chellos, full orchestra, etc.). That is, if I say "this strings sound great", it only means that it's great for my type of music.

When I look after specific instrument sound (that I can't find on keyboard or in some expansion pack), then I usually first check Soundfonts 4U, where we can also find few strings soundfonts. Of course, that means we need to create a custom voice out of soundfont.

Bogdan

[Lee Batchelor]

Interesting comment about the memory requirements for the CFX Piano, Bogdan.

Before owning the Genos, I used a Tyros 5 and Nord Electro 5D. For years the Nord pianos were considered dismal, so Nord did something about it. They dedicated a full GB just for their piano sample. At the time, that was a lot of memory just for one voice. However, the new piano voice was amazing!

Yamaha engineers could easily give us the best string samples available by adding more memory. They simply choose not to because the marketing people wouldn't agree. In general, the Genos has excellent voices but Yamaha has packed a little bit of everything into the Genos, which leaves gaps like the one Tommy writes about.

Tommy, it sounds like you're a home player. When you play, you don't have the ambient noise we stage players experience. On stage, the subtle distortions are easily masked. At home, you can hear a pin drop between songs. You want high end string sounds with no distortion and everyone here agrees. Sadly, you must invest in high-end VST sounds to achieve that. I find the same issue exists with the pianos, electric pianos, and B3 organs. For home playing they are barely "okay." They are suitable for the live stage because the missing finer details are masked against all kinds of other sounds.

[mikf]

Any kind of recording is much more demanding than any kind of live playing, home or in public. It's not just sound quality, noises or distortions, .....in live playing you can even make small mistakes and they may hardly be noticed, because they are gone, in the past, instantly.
But even the smallest mistake on a recording has to be fixed, because within a couple of listens it becomes the only thing you can hear. It's why studio work is so demanding, and perfectionists do 30 takes, then still make cut and paste corrections.
And as far as instrument simulations are concerned, even the most perfectly made sample is never everything to everyone. Because who plays it, how it is played and the sampling set up all have an effect on the end result. It is a simulation, never the real thing.
That's the beauty of the real thing, good players can all get different sounds from an instrument. Two excellent samples of the same instrument can be quite different, and appeal to different people. It's choice and preference as much as good and bad.
I have never noticed the distortions being talked about here on my CVP, but I'm sure that is not because the CVP samples are better. It's just because I have never noticed. Maybe I would if I was trying to make certain type of perfect recordings. But it has never mattered to me so far, or indeed most of us..because I'm not Oscar Peterson or Frank Sinatra and perfection isn't a word that can be applied to my recordings.
While obviously annoying to Tommy, maybe even a deal breaker, my own view is that here we have $6000 instrument that packs dozens and dozens of pretty decent instrument simulations. So now someone has  found an imperfection in a few........is that really a big deal?

Mike

[Lee Batchelor]

Any kind of recording is much more demanding than any kind of live playing, home or in public.

You got that right, Mike!! If you really want to find out how good or bad you sound, forget practicing for hours. Take five minutes to record yourself and play it back. You'll soon clean up your mess.

The most common error I find people make is "rushing the chord or melody line." This applies to ALL musicians. I took lessons from Shania Twain's piano player. He said, "When you rush any piano passage you sound like an amateur. It's when you play ever so slightly behind the beat is when you start to sound like a pro."

Also, we should not confuse sample quality with sample defect. Tommy is talking about the latter. You can have a midstream sample that does not distort OR you can have a high quality sample with some keys that do distort. The CFX Grand on the Genos has a few keys that have a distortion. Very annoying for home players. Not so for stage players because we don't hear it, mostly thanks to drummers and guitarists 🤣!

[BogdanH]

hi Lee,

Interesting comment about the memory requirements for the CFX Piano, Bogdan.

Maybe I should clarify why such huge size (btw. I'm talking about Garritan CFX VST library)....
The library contains three different piano recordings and the difference between them is where microphones were positioned. For example, there's "classic" library, where microphones were positioned at the side of piano (that is, where the audience would hear it). And then there is "player" library, where microphones had player position (that is, listening from keybed side). Obviously, these libraries have slightly different sound.
When we decide to use the VST (or make a custom voice), we need to decide which sound we prefer and so we only use samples from that particular set of samples (i.e. classical). That way we already reduce the size of the library to one third. Means, if we decide for "classic" sound, then it will have size of only 40GB.
Each set contains separately recorded samples of ambience (microphones had certain distance from piano), separately recorded samples of hammers and separately recorded sustain samples. And yes, there are also separately recorded samples at closed lid, at partially open lid and fully open lid.
When used as VST, we can anytime mix all that to our desire, but that's not possible if we wish to create custom voice -because a single voice can only have certain sound (which we can modify later on keyboard). Anyway, after we eliminate certain samples (i.e. ambient, closed lid, etc.), we end up with say, 8GB. That's still too much for creating a voice..
Notes are recorded at 8-10 velocities, which is great, but we don't really need such precision -because it takes a lot of memory and a lot of time to create a voice, and so four velocities is just fine. By reducing amount of velocities, we further reduced the size.
Now.. each velocity contains 80+ samples (almost every note has separate sample). We don't need that many.. if we use four samples per octave, it should be just fine: 4 x 6octave =24 samples only.
Depending on the tone, each sample has length of 10-30seconds. I found out that in most cases that's very generous because almost a half of sample contains silence -that's especially true for higher notes (C3 and higher). Solution: cut the samples to desired length.
To make (too long story) shorter: we end up with a voice that has size of about 50MB. We need to keep in mind, that this voice represents only one particular setting in VST!

PS: Genos owners already have built-in CFX piano and so I don't think it's worth to bother with all that. Strings? Well, now you have an impression of how complex it can be.

I hope that was interesting for some of you,

@Mike -fully agree with you.

@Lee

...Also, we should not confuse sample quality with sample defect.
...

-YES! As is the case for Saxophone voice I complained about: it sounds just great, but C4-G4 issue is something I can't forgive.

Bogdan

[Lee Batchelor]

Thanks for the explanation, Bogdan. I knew about the multiple mic positions and hammer sampling, open lid, closed lid, etc...but I didn't know how they processed it all after the sampling was complete. I'm sure your explanation is just the tip of the iceberg. The engineering gurus may have more to add 🤣. Thanks again...

- Lee

[mikf]

Lee, He is so right about playing slightly behind the beat, especially in middle of the road music, standards, etc. maybe less so in heavy beat music like rock. For some reason being unhurried seems to exude confidence in playing and makes listeners relax. It's a very small delay though, almost a feeling rather than something measurable.
Mike

[andyg]

Yes, just behind the beat works in modern types of music and yes, it's got to be almost unnoticeable. But I don't think Herr Bach would appreciate it done too much to his music!

FWIW, I used to review virtual instruments as part of my job. East/West, Garritan. Miroslav, Kirk Hunter etc etc. I too have pianos that take up many Gb of space on several hard drives.

Strings were always going to be hard to sample, choirs too. I was sent one set of strings (I'll not mention the name but it was well known at the time) that retailed at around £1,000. Overall, wonderful, but if you listened uber-critically, you could still hear some quantisation artefacts and noise. VST sampling has improved since then, but as has been said, keyboards don't have the quality of samples that you find in a good VST.

And remember that noise is cumulative. If there's noise in a sample when you play C, adding E and G will treble the noise - numerically at least. I was once set the task of discovering if you could produce better ensemble strings by layering many solo strings together. I had plenty of those, but the time I'd got to half a dozen violins, the digital artefacts and noise was very noticeable.

And remember also that adjacent notes on a keyboard may well share the same sample, being replayed at a slightly different speed, with the possibility of phasing artefacts rearing their ugly heads. Oh, the fun you can have in R&D, using S&S to try to avoid that!

On the whole, today's keyboards, regardless of make, do a darned good job at what they're intended to do - being played live, and in the vast majority of cases by amateur home players. Put them into a studio type situation and ask them to act as sound sources and you'll show up weaknesses, but it's taking them out of their comfort zone and making them do something that they weren't really meant to do.

[Lee Batchelor]

All great discussions, folks. It goes to show that sometimes it's healthy to drift outside the main top a wee bit.

[Amwilburn]

There is no defense of the Yamaha keyboard.
You won't find SEK 50,000 under the bed.
When it also applies to the previous models with the same problem and they continue without improving.
No Yamaha has fallen in my eyes.
Sorry, but there won't be any more Yamaha keyboards for me
Tommy

You should also be aware that the "spittle" sax also shows up on the Korgs, incl the PA5x. One of the consequences of using samples, is you sometimes get *all* the real sounds, even the ones you don't want. However, the solos strings on the Yamahas should *never* have included the piezo pickup 'thwap' sound from the bow hitting the strings quickly. They should've just used a pair of condenser mics for that (or both, and envelope filter or even record *only* the tap and subtract the signal) and I've absolutely no idea why they didn't.

Well, it has been mentioned few times before, that strings on Yamaha's aren't really that good... it only comes down to what makes them "not that good". I quite like listening Classical music (when I'm in the mood) and in my opinion, strings are far from good. They're usable as background orchestra, but not as leading instruments -just my opinion.
I only have Yamaha and so I can't say if other brands have better strings.

Bogdan

This is an interesting question, because to *me* the orchestral ensemble strings are *far and away* the best on Yamaha... for action soundtracks (2nd best I'd ever seen was the Roland SuperNatural string ensemble from Jupiter 50/80. The sound wasn't as *real*, but the way the strings played were *better*. It's like a much better version of ensemble mode, actually.

When you get to the 'romantic film' kind of slow strings? Yamaha had pretty good ones ages ago (slow strings under Legacy) but other brands *were* better for that. (The new Cinema strings on G2 is also really good for that, btw). For solo strings? this is a *much* tougher call. The single slow cello introduced in CVP209 (possibly even before) was oddly never put into Tyros/Genos, so it took until Tyros 5 to even have a decent bowed single cello; the Korgs and the Rolands had always focused on solo string instruments over the ensemble ones: so for a *long* time they've both had really good solo string instruments, and really good slow romantic ensemble strings. Fast, orchestral "action film" strings were great on the Jupiter 50/80, but not much else; even the Korg Pa5x has decent Movie strings if the fast bowed ilk, but the higher notes sound *really* synthetic. Yamaha's sound *real* almost all across the keyboard... but again, that's only for the fast bowed strings.

@Lee-YES! As is the case for Saxophone voice I complained about: it sounds just great, but C4-G4 issue is something I can't forgive.

Bogdan

Oh, btw, not only were the "splittle" sounds from legato pianissimo sax a complaint you had on Yamaha, I should point on they're on the PA5x as well. *and* I checked, they're on Montage /Modx as well (I haven't checked on the new Montage M, because while we've had stock, we've had no *spare* stock... Montage M keeps selling out before we can put them out on display. So I've yet had a chance to actually play one!). But like Andy said, it's not a defect in recording, it's an annoying choice of which sample to use (most likely they didn't notice the spit take while recording... And yes, it's avoidable, as the condensation that causes it in a sax takes a while to form. But they'd have to literally wait to clear the condensation between takes, quite time consuming. And you obviously never noticed it live; but I've heard it happen live (a lot... you'd be surprised the number of sax players we get). The sample engineers should've known about it though. Just like the stupid piezo pickup on solo string instruments... just use a paid of condensers like you do for an ensemble! The former issue is a musical instrument issue; the 2nd is purely from a recording standpoint. The "thwap" doesn't happen in real life *unless* you're a bridge pickup into an amplifier, which really only happens when you're busking!

Sometimes, you don't want real :p I had a customer complain that the piano sound on their keyboard was too bright, unlike a real piano (I then took them over to the real grand that it was sampled from, and they were *stunned* to hear it was just as bright in reality. So I switched his keyboard the the GM/XG piano, and he loved it :p Even though that was a synthesized piano on his instrument (this was over a decade ago)

Mark

[BogdanH]

..
Sometimes, you don't want real :p I had a customer complain that the piano sound on their keyboard was too bright, unlike a real piano (I then took them over to the real grand that it was sampled from, and they were *stunned* to hear it was just as bright in reality....

-you've hit the nail! 
The thing is, we want perfect and real at the same time: we want string sound without a bow, we want piano without any mechanical noise, we want singer without hearing singer taking the breath, sax without spitting distortion (in certain notes), etc. But sometimes subtle side noise is desired, i.e. slide finger along steel guitar string -but if it's too much, then it becomes very annoying.
Imagine string orchestra voice where you can also hear audience coughing when you play certain notes... hey, is realistic 
Anyway, in my opinion, such voices (containing real noise or distortion) should only be an option that user can choose if he wish.

You also gave good example with that customer you had (about how piano should sound) -many times we don't want real.. it should sound nice instead 
Maybe that's one of the reasons why we sometimes disagree about sound quality of certain voices 

Greetings,
Bogdan

[mikf]

And people also want perfect for relatively low cost!
There are many who post how they are dissatisfied that an instrument costing "$5000" is not completely perfect. This is unrealistic. There are about 2000 voices and 600 styles on these instruments, taking into account all the other features, keyboard, controls, this is not even a dollar a voice. The only reason it can be so low cost is because Yamaha can leverage historical models and other synth products.
  A Steinway can cost $100,000, a quality sax is about the same as a Genos, and even a quality acoustic guitar is many thousands. I am not saying that Yamaha shouldn't be trying to make it as good as they can, and I know there are many for whom a $5000 purchase is a big deal, but the fact is these instruments - and I include competitors products like Korg in this - are phenomenal value for money compared to many other musical instruments I have owned.
Mike

[Amwilburn]

.
Imagine string orchestra voice where you can also hear audience coughing when you play certain notes... hey, is realistic 
Anyway, in my opinion, such voices (containing real noise or distortion) should only be an option that user can choose if he wish.

Greetings,
Bogdan

  hahahah I can just see it now: they'll be touting the recording of the sound of the audience talking as a *realistic feature*.
"But they were present at the recording, so it's real!"

[Lee Batchelor]

One thing that has not been mentioned in this ever growing thread is, "how we use the instruments in question." I've heard very expensive VST voices that, when played by themselves, sound dismal and cheap. As soon as they are added to a mix, voila! They sound exactly like the real thing.

Lots of expert players perform as soloists but these keyboards were meant mostly for ensemble playing. Anyone who expects to play a classical tune with a Genos CFX Grand through top of the line speakers and have it sound like an actual Yamaha grand, is only fooling themselves. I won't happen.

So, how do these voices sound in the mix? Pretty darn good I'd say. And yes, I complain about paying $7,000 for substandard pianos, e.pianos, and B3 organs but that's just me. In a band mix or duet with styles, people are blown away by the sounds.

[BogdanH]

..."how we use the instruments in question." I've heard very expensive VST voices that, when played by themselves, sound dismal and cheap. As soon as they are added to a mix, voila! They sound exactly like the real thing....

I think that's also true for many keyboard voices, because

...these keyboards were meant mostly for ensemble playing.

Anyone who expects to play a Classical tune with a Genos CFX Grand through top of the line speakers and have it sound like an actual Yamaha grand, is only fooling themselves. I won't happen.

hmmm.. is probably true.
But there's also another reason why it doesn't sound real (even if voice is perfect):
depending on classical piano piece, average real piano loudness goes way above 90dB (yes, piano is loud instrument!). And how loud we play at home? Relative quiet usually and so no wonder we can't hear all sound details (not to mention feel the resonance of the piano).
We want to hear realistic sound? Then we also need to listen at realistic loudness.

Bogdan

[Lee Batchelor]

But there's also another reason why it doesn't sound real (even if voice is perfect):
depending on Classical piano piece, average real piano loudness goes way above 90dB (yes, piano is loud instrument!). And how loud we play at home? Relative quiet usually and so no wonder we can't hear all sound details (not to mention feel the resonance of the piano).
We want to hear realistic sound? Then we also need to listen at realistic loudness.

Very true, Bogdan. The concept of dynamics has not been mentioned and has a huge effect on the overall sound.

[andyg]

Sounds that are just too realistic can be a problem - it has been ever since sampling 'came of age' and we got a good number of samples spread across the keyboard, coupled with the sampling itself getting better.

Ironic, people ask for ever more realistic voices, without realising that half the 'realism' is in how the voice is played and used in the piece. Something that has to be taught in order for people to be able to phrase and articulate. Alas, it seldom is taught and with the vast number of self taught people....

I can remember some makes of organs around the turn of the century deliberately not having ultra realistic sounds. If they had such sounds they would have never blended with the more traditional home organ sounds, instead standing out like sore thumbs!

Roland's strings have been mentioned. I'm lucky enough to have the Atelier AT900 Platinum, the last organ Roland ever produced. It's not perfect, no instrument ever is, but I have to say that it has ensemble strings to die for! You can play them one at a time but when you start mixing them, the possibilities are amazing. But.... they don't have a good solo violin or cello. You can, however, fake them very well with a cunning mix of two voices.

[Amwilburn]

Oh, I wasn't even thinking of the organs... in Canada, none of the major companies has brought organs in since around the turn of the millenium; I'd forgotten Roland even used to have an organ arm! I guess they must still exist in the States? Organs are still a *big* deal in Asia, hence YEM support for the StageA

In fact, if you search organs on Roland Canada, the web page still comes up showing Atelier & Classic, but if you click into those, nothing :p

https://www.roland.com/ca/categories/organs/

Mark

[mikf]

Andy has it right, how you play is a huge part of how an instrument sounds.
But there are differences and the sample/voice may not be the biggest difference. I have two clavinovas and two quality grand pianos, so I am able to compare pianos side by side in real situations. The voices are really not the big issue. Clavinova performs pretty well, but the three big differences with quality acoustic pianos are feel, resonance and sustain.
The  first two are understandable, nothing can compare with the feel of a real high quality piano. And the room resonance and sheer presence of a quality Grand can be stunning. I don't know if any digital can get there.
But where I feel a little bit cheated by most digitals including the CVP is on sustain. I am a big sustain user when playing piano, because my playing style emphasizes bass, which I typically sustain over the other notes and chords,  and I don't know why that can't be better on digitals.  And it's not just the Clavinova, which is pretty good by typical digital piano standards. I have never played any digital piano where I really was impressed by the sustain.
Mike