Rates of improvement?

[Tusker]

Does Moore's law hold for keyboards and synths? Would you expect rates of improvement to stay constant?

 

Just wondering what your experience is. I've thought that rompler/workstations haven't advanced recently as much as software sample solutions have. That seems to be a step-function improvement. And it's not fully played out yet by all indications.

 

But for hardware I am expecting declining rates of improvement as in any maturing technology. Should I?

 

(BTW, I am quite happy with the industry. Please don't take this as a dig. Just wanting a non-politicized discussion about technology.)

 

But I've also wondered if I am becoming a curmudgeon (liking what I know and knowing what I like). And whether we just become sticky with time. (Stuck to our stuff.) Are you stuck?

 

Thanks for your comments,

 

Jerry

[soundscape]

Moore's Law doesn't apply to software, except where hardware speed improvements enable a practical implementation.

 

Just wondering what your experience is. I've thought that rompler/workstations haven't advanced recently as much as software sample solutions have.

Many of these may appear to be better, but I think you have to check if they are actually "better." http://www.discodsp.com/highlife/aliasing/ shows the aliasing issues of many software samplers, for example. (I wish someone would post more a more up-to-date test.)

[lerber3]

Moore's law basically applies to computing capacity: size, bandwidth, speed. The law proposes that these will double every 18 months.

 

Software and hardware can't really be measured in this way... at some point, a piece of software or hardware is 'good enough' at fulfilling its purpose. It is a bit difficult to think of it being 'twice as good' without also changing the purpose.

 

Alternatively, we could consider that perfection exists and each generation of solution is twice as perfect... the improvements have to get smaller & smaller. Probably only the most curmudgeonly among us would notice after a while.

 

Lerb

 

Maybe you will be twice as curmudgeonly after this answer.

[ProfD]

I think over time, we either a) get stuck to our stuff or b)comfortable with something along the same lines.

 

I'm in the camp of stuck to my stuff. I keep listening to the MotifES8 and S90ES but don't hear enough of a difference to upgrade from my S80.

 

I remember when the difference in sound and features between keyboards by manufacturer was substantial enough to justify buying one of each brand. Not today.

 

It appears that most boards within the same market i.e. DP, workstation, VA, etc., all cover the same bases.

 

In fact, it appears they have thrown everything into a board nowadays. There seems to be enough parity to the point where it really comes down to personal preference.

[soundscape]

Originally posted by lerber3:

Alternatively, we could consider that perfection exists and each generation of solution is twice as perfect... the improvements have to get smaller & smaller. Probably only the most curmudgeonly among us would notice after a while.

There is no reason to believe it always gets "BETTER"... (among digital synths) the JD-800 is still a sought after instrument, and as noted in another thread, the E-mu Morpheus had 197 filters while the current E-mu (Proteus-X2/Emulator-X2) has has 50 filters.

 

Similarly, the Sound on Sound review of the Kurzweil K2661 says: "Make no mistake; in terms of the quality and breadth of their sounds, I believe that Kurzweil are still at the top of the pile, even after all these years. The sounds just seem to leap out of the speakers and caress your ears." and "The K2661 inherits the formidable sample-editing capabilities of the K2-series and it is fair to say would still put a fair number of dedicated samplers hardware and software to shame. VAST itself is more than capable of the real-time enhancing or mangling of sample material beyond all recognition."

 

The thing is that at one point, a relatively small increase in processing power and memory brought with it huge gains. For example, the Korg M1 didn't have enough Wave ROM for a "real" sounding piano. Up the memory a bit, you can get significant changes in the sound. However, once you get to 50Mb for a piano sample and you up it to 1000Mb, it's not going to sound 20x better; in fact, the improvements may only be demonstrable under certain circumstances (e.g., unlooped samples only on longer notes.) Meanwhile, if the piano sampled, the care taken in sampling it, the programming, and the synth engine, are all substandard, then the quality will be inferior. (Or even atrocious)

 

OTOH, as noted in another thread, we see technologies such as the General Music "FADE" filters which attempt to filter the partials so that they change just like they would on a real piano as the keys are striked harder, and other "physical modelling" technologies... I'd expect Moore's Law helps with the processing necessary.

[Finale]

My view is that we are extremely spoiled today, and that many don't realize this. For example, compare what you can buy today in terms of synths even for $50. In the 70s, all I could buy was a Magnus organ, one of those reed instruments that sounded like an cheap accordion. Today, you can buy a toy-keyboard with dozens of different sounds, auto-accompaniment, etc. Of course, no comparison with pro stuff, but you get the idea on how the technology has improved.

 

A major problem today is that people want complete solutions without any effort, right away. In a sense, we live in this kind of mentality since years in many domains. People want incredible-sounding digital piano factory presets, but they don't want to work more than five minutes to adjust it to their taste. How many people do we read every week here complaining about this sound, this brand, this action? But how many truly tried to adjust those sounds or velocity maps to get closer to their criterias?

 

There is certainly room for improvement in terms of interface (more than compared to sound quality), but I believe that many should focus instead on how to use their tools more efficiently. And this includes to read the manual (RTFM). Or, if one realizes an instrument can't do what was expected, find another that offers more programmability.

 

My conclusion, since HAL-9000 doesn't exist yet, don't wait for the technology to do everything for you. There's still a lot of work to be done on the "human side". And maybe it's better that way. Remember what happened ultimately with HAL.

[ProfD]

Originally posted by Cydonia:

There is certainly room for improvement in terms of interface (more than compared to sound quality), but I believe that many should focus instead on how to use their tools more efficiently. And this includes to read the manual (RTFM).

+1

 

Cydonia, you know this would put some companies out of business, right?

 

Planned obsolescence is the game. Release new product with minimal gains or a boat load of features some will never use.

 

I see it all of the time and I'm still amazed by folks who upgrade equipment (software, synths, cars, etc.)when they haven't even begun to probe underneath the hood of the one they bought last year, week, etc.

[cnegrad]

But at this point, rompler/workstation manufacturers are up against a bit of a brick wall, if you ask me. Other than synthesis engines, rompler sampled sounds can't hold a candle to the large sample libraries now available. So they mostly sound inferior, with the possible exception of synthesis engines. Until memory becomes dirt cheap, the hardware romplers will always sound inferior, unless a quantum jump in physical modeling happens soon.

[JeffLearman]

I expect to see big improvements in ROMpler pianos in the next few years, because IMHO there is a huge difference in potential quality between a 20MB piano and a 200MB piano.

 

50 MB is cutting it close. That's even less than my 70MB Rhodes, with only 5 velocity layers, unlooped samples up to 25 seconds, and every 4th white note sampled.

 

Here are the flaws with current memory limitations in ROMplers.

 

1) Samples are far too short. This causes them to go "static" as they enter the loop. I can hear this immediately at 1000 paces, and I'm not any golden-eared type. When you're used to a piano that's "alive" for several seconds and hear a ROMpler where most samples are less than a second, you miss it. This is especially true for stereo pianos, where the image gets frozen, unlike either a real piano or a longer-sampled digital. IMHO, 4 seconds is the minimum, and if I were the sound designer for a major ROMpler, that's probably the length I'd use (using any left over memory for more notes or more layers).

 

2) Most ROMplers use 3 layers, and it WAY ain't enough. Few ROMplers sound really good for sensitive, subtle playing in the ppp to p range. They're fine for rock and roll, man. I think 8 layers is enough. With careful velocity sensitive EQ you can make 3 layers work a lot better than most ROMplers, but apply that same care to 8 layers and it makes a very obvious difference when you play.

 

3) Every 4th white note means that I frequently play chords where the same sample is used for more than one note. Someone here first brought that issue to my attention, and now that I'm aware of it, I agree wholeheartedly. It's not as subtle a difference as I'd like. Of course, to completely avoid this we'd need to sample every note. I think that every 3rd semitone is a reasonable compromise, and every other semitone would be even better.

 

Now, keep in mind I'm talking about a Rhodes here, and Rhodes is FAR easier to sample well than piano, because the sound is much less complex. (OK, there's more timber change in the high velocities of low notes, but not too much more than a big grand piano.)

 

However, as someone mentioned on another thread, more memory does not equal better sound. It just gives the sound designer more rope, with which he can do acrobatics or hang himself -- and the latter is far easier. Sampling skill is paramount.

 

However, I expect to see a pretty serious levelling off in the foreseeable future, and improvements will be more in the nature of cost/performance ratio and better user interfaces.

 

For example, there are 3 or 4 very good Hammond sims; I'd be satisfied with any of them. None of them quite does the key click (though I haven't tried the Suzuki/Hammond yet). Once they get that down, and with a few improvments on Leslie simulation, what's left to improve? Not much.

 

Another improvement will be putting more excellent sounds in the same package.

 

Yamaha needs a good Hammond sim! That's the main area where they're a 2nd class (or 3rd class) operator, and the main reason I'd prefer an RD700sx over an S90ES. (There are good reasons to prefer the S90ES, of course.)

[ITGITC]

Originally posted by Cydonia:

My view is that we are extremely spoiled today...

Yeah. Ummmmmmmmmm, OK.

 

So where's the new Kurz PC1se?

 

C/Bb7

[Finale]

Originally posted by Is There Gas in the Car?:

Yeah. Ummmmmmmmmm, OK.

 

So where's the new Kurz PC1se?

Well, it's available now, one guy recently gave a few comments on it here. I thought you were happy with your PC2X.

 

Now go read your manual before complaining you need a PC3X next week.

[ITGITC]

Originally posted by Cydonia:

Now go read your manual before complaining you need a PC3X next week.

[DanS]

Originally posted by Cydonia:

My view is that we are extremely spoiled today, and that many don't realize this.

Couldn't agree more.

[soundscape]

Originally posted by cnegrad:

But at this point, rompler/workstation manufacturers are up against a bit of a brick wall, if you ask me. Other than synthesis engines, rompler sampled sounds can't hold a candle to the large sample libraries now available. So they mostly sound inferior, with the possible exception of synthesis engines. Until memory becomes dirt cheap, the hardware romplers will always sound inferior, unless a quantum jump in physical modeling happens soon.

In a sample and synthesis engine, the point is that the synthesis IS (or at least can be) a large part of the sound, so it's not a trivial consideration. (BTW, your post implies that synthesized sounds "can't hold a candle" to acoustic (or electro-acoustic) sounds.)

 

Also as has been mentioned, the GEM "FADE" filter is supposed to alter the piano's harmonics just like a real piano; also the VAST programming in the Keysolutions Piano is discussed in the message: http://www.musicplayer.com/ubb/ultimatebb.php?/ubb/get_topic/f/18/t/020532/p/2.html#000039

[Tusker]

Originally posted by lerber3:

Alternatively, we could consider that perfection exists and each generation of solution is twice as perfect... the improvements have to get smaller & smaller. Probably only the most curmudgeonly among us would notice after a while.

I agree. The limits of perception are a very big key to this question. For samples (especially in live use) I think we got pretty good 5-10 years ago. At least that's my perception. Which is why I am asking this question.

 

To continue comparing us to the computing world, computers came up with all kinds of high end applications (color, games, video, etc.) that "had" to be placed on ones desktop. By contrast most synthesists I know (myself included) are ok with black and white screens if the sound is good.

 

When it comes to perception, I also think there is a "bench test" mentality that is similar to computers. You can run mandelbrot equations to test computers, but most of us want to do wordprocessing and play some games. Similarly for aliasing tests ... all digital synths alias at some point. If you really want a saw wave to sound clear 4 octaves about middle c, just use analog and you are done.

 

However that said ... would you expect huge gains in power over the next decade? If so, what on earth are we going to do with it?

 

Jerry

[Tusker]

Originally posted by learjeff:

I expect to see big improvements in ROMpler pianos in the next few years, because IMHO there is a huge difference in potential quality between a 20MB piano and a 200MB piano.

That sounds like a real need that manufacturers could address. It's not a need for me (I use an acoustic piano for recording, and for live stuff, the romplers are ok.) If I was playing in a jazz trio or solo piano, I might care though.

 

I think it's important to have a need to improve, don't you?

 

Jerry

[Tusker]

Originally posted by DanS:

quote:

---

Originally posted by Cydonia:

My view is that we are extremely spoiled today, and that many don't realize this.

Couldn't agree more.

[Tusker]

Originally posted by soundscape:

Originally posted by cnegrad:

But at this point, rompler/workstation manufacturers are up against a bit of a brick wall, if you ask me. Other than synthesis engines, rompler sampled sounds can't hold a candle to the large sample libraries now available. So they mostly sound inferior, with the possible exception of synthesis engines. Until memory becomes dirt cheap, the hardware romplers will always sound inferior, unless a quantum jump in physical modeling happens soon.

In a sample and synthesis engine, the point is that the synthesis IS (or at least can be) a large part of the sound, so it's not a trivial consideration. (BTW, your post implies that synthesized sounds "can't hold a candle" to acoustic (or electro-acoustic) sounds.)

 

Also as has been mentioned, the GEM "FADE" filter is supposed to alter the piano's harmonics just like a real piano; also the VAST programming in the Keysolutions Piano is discussed in the message: http://www.musicplayer.com/ubb/ultimatebb.php?/ubb/get_topic/f/18/t/020532/p/2.html#000039

There's some interesting information on that thread. Thanks. I too have been turning to synthesis in lieu of larger samples. Elhardt does a lot of cool stuff in this area. Here is his synthesized piano done on a Arturia Modular:

 

http://home.att.net/~synth6/Arturia_MMV_Piano.mp3

 

I find these kinds of techniques to be part of the improvement cycle. It's not just the manufacturers, the hardware and the software. It's us as players being ingenious in our use of synth resources. While I don't do emulations on the Nord (I'm no Elhardt) I have enjoyed getting controllable gutty and reedy qualities similar to cello and large bore woodwinds ... just by tuning the filters. Things you don't get from a bigger sample. Best,

 

Jerry

[cnegrad]

your post implies that synthesized sounds "can't hold a candle" to acoustic (or electro-acoustic) sounds.

Don't know where you got that from. Please don't read into any "hidden meanings". I specified that when comparing _purely sampled sounds in romplers_ (without any synthesis involved) to those in large sample libraries, the large sample libs will come out on top. And as a result, the _purely sampled sounds_ in a rompler will make the rompler sound poor by comparison. In my original post I went out of my way to exclude synthesis from my comments.

For samples (especially in live use) I think we got pretty good 5-10 years ago.

I'm also not thrilled about the fact that we've grown accostomed to the new mindset of, "well it's good enough for live use". If I'm going to spend thousands on a hardware board, I want it's sampled sounds to be good enough for the studio.

[Tusker]

Originally posted by cnegrad:

I'm also not thrilled about the fact that we've grown accostomed to the new mindset of, "well it's good enough for live use". If I'm going to spend thousands on a hardware board, I want it's sampled sounds to be good enough for the studio.

It used to be that way. There's a bunch of 1990s CD's with rompler pianos, string beds etc. Even the Korg M1 piano is out there on a thousand CD's.

[cnegrad]

But back then those sounds were all that was available, unless you were one of the lucky ones who had access to a Fairlight or Synclavier. Now, original recordings use the high-end libraries, and the romplers just can't keep up, unless you're using a Neko or a Receptor. It's tough now to sit down at my gig at my Roland rompler after playing giga sized high end libraries in the studio all week.

[soundscape]

Originally posted by Tusker:

When it comes to perception, I also think there is a "bench test" mentality that is similar to computers.

Hmm, I'd say there's actually a complete lack of measurements and technical info.

 

Originally posted by Tusker:

ou can run mandelbrot equations to test computers, but most of us want to do wordprocessing and play some games. Similarly for aliasing tests ... all digital synths alias at some point. If you really want a saw wave to sound clear 4 octaves about middle c, just use analog and you are done.

Yet, in this aliasing test, the SOUNDBLASTER LIVE! gets better results than many software samplers (EXS-24, Halion and Reason):

 

http://jeskola.net/xs1/test

 

I wish there were a more up to date test posted, but--I'm quite sure when this was posted these samplers were described as the "NEW FRONTIER."

[soundscape]

BTW... here's a clip from "The Sims 2"... this is garbage quality and sounds like software to me.

 

http://download.yousendit.com/3ECAE2B948AF4E5C

[Dan Phillips]

Originally posted by soundscape:

Yet, in this aliasing test, the SOUNDBLASTER LIVE! gets better results than many software samplers (EXS-24, Halion and Reason):

Hardware by Emu, I believe.

 

- Dan

[Dan Phillips]

Originally posted by soundscape:

Yet, in this aliasing test, the SOUNDBLASTER LIVE! gets better results than many software samplers (EXS-24, Halion and Reason):

SBLIve featured hardware from Emu, who had very good aliasing performance in their EIII and EIV samplers (and presumably in Emulator X, as well).

 

- Dan

[ITGITC]

Originally posted by cnegrad:

 

It's tough now to sit down at my gig at my Roland rompler after playing giga sized high end libraries in the studio all week.

Originally posted by Cydonia:

 

My view is that we are extremely spoiled today...

 

 

Hi cnegrad

 

Luv ya... mean it.

 

Is There Gas In The Car?

[cnegrad]

Point taken.

[soundscape]

Originally posted by Dan Phillips:

Originally posted by soundscape:

Yet, in this aliasing test, the SOUNDBLASTER LIVE! gets better results than many software samplers (EXS-24, Halion and Reason):

SBLIve featured hardware from Emu, who had very good aliasing performance in their EIII and EIV samplers (and presumably in Emulator X, as well).

Exactly. It uses E-mu's 8-point interpolation. (IIRC the Emulator-X is higher quality, but I can't find any information right now.)

 

More than 12 years ago, the SoundBlaster AWE-32 "consumer" soundcard had an E-mu chip (the E-mu 8000) and could take up to 28Mb on-board RAM (SIMM slots onboard) for loading SoundFonts.

[kanker.]

All I know is that Moore's Law is one of the most woefully misunderstood concepts I've ever run into.

[ITGITC]

Originally posted by cnegrad:

Point taken.

You know I'm just messin' wit ya, cnegrad.

 

We are all spoiled - ain't technology great?

 

And it's true - there's a LOT more I can learn about my synths, but it's easy to dream about how buying a new one can make everything better.

 

I'm up to page six on the other thread. Gotta go pick up where I left off.

 

C/Bb7

 

Take care,

 

Tom