Is Sampling Becoming Obsolete?

Very, very interesting video about Pianoteq compared to NI libraries.
Physical modeling vs. sample library.

3 Likes

Wow that actually seems nice, will have to listen through my brothers studio monitors later! Thanks for posting

1 Like

Maybe in this instance but as a whole, no. Gotta say the title is a little bit of click bait.

14 Likes

Pianoteq does sound really, really nice. I agree with @panelist, though. Title is misleading. I love PM synthesis and am also a bowed and plucked string player. I have not yet heard a convincing model of a lot of stringed instruments (obviously except keyboard strings). Especially bowed instruments with sympathetic strings played in a non-Western style.

2 Likes

I agree.
But he does not say sampling has become obsolete. But based on this example, he thinks that physical modeling has the potential to replace it in future.
As far as technique has come, the question might be justified.
This piano sounds really impressive.

1 Like

Here my few cents on this ā€¦

Once upon a time, I believed it possible to create an authentic sax sound with a modular synth. The reason, why I tried this was, I had no saxophone. I failed. Later I tried a Yamaha VL70-m, which is a physical modelling synth. Way better, but compared to real saxophon players, I missed very much.

This said ā€¦ I got a sax, learned how to play and think that I understand now, why this synthesizing or physical modelling could, as simple as is, not generate the real tone.

A sax player can do so many things with his breath, his tongue, his voice, his lips, and the forming of his mouth, all of this simultaneously, and with different nuance, which a non-sax player just doesnā€™t know about in the first place, rather then knowing how to simulate this with a non-sax, or even a sax-like interface.

I think thatā€™s true for many traditional instruments too. I takes the original instrument and an educated musician to create those authentic tones, we love. Faking might become better and better with time and improved technology, but at the end, it often could be a quicker and better result, just to employ the ā€œreal thingā€ :wink:

4 Likes

I would certainly agree, that itā€™s nearly impossible to completely fake real musicians!
But this is about Physical modeling vs Sample libraries, not musicians.
Sample libraries are played with midi keyboards, or even midi programmed in a DAW.
If they would manage to get PM as good sounding and behaving as costly recorded and edited top notch Kontakt libraries, that would be amazing. No tons of gigabytes, fast loading etcā€¦
Itā€™s really really impressive what some people can do with sample libraries.

3 Likes

Thatā€™s exactly, what I think will not be delivered for the next time. The tone is dependent on how it has been created by a musician. There are sampled instruments, which are not only tones, but played tones or phrases by musicians. Those sounds are sometimes undistinguished from a real player in the studio, because they have been created by a real player, who had the skills and has been recorded.

Many classical instruments allow for much more modulations rather than a keyboard, even with some additional wheels or joy-sticks :wink:

1 Like

yes, you are right. Also depends on the instrument.
At least for piano, they seem to come close.

Indeed.
And while it is interesting that research has reached such a state, Iā€™m more interested about expanding the models outside of the perimeter of the simulation. Grey zones are where lie unknown sounds :wink:

Oh, and imagine LFOs on all these parameters !!!
:smiley:

7 Likes

That would indeed be cool! I would love that!

Not to mention, with parameter control, you could automate the sounds to warped into oblivion- the sounds of instruments could be disintegrated into oblivion-which would be really cool

1 Like

This sax player agrees. :slight_smile:

3 Likes

but why does everything HAVE TO be artificial? I donā€™t understand this obsession

Even if piano physical modelling will get close to perfection, its just as laughable an idea as claiming AI will match human intelligence just cause AI can win a human at chess or at go IMO. Simulating reality will always succeed better in some areas of life and worse on othersā€¦ Samples and sampling will stand the test of time. Why choose anyway? The more options, the merrier.

1 Like

MPE controllers such as the Roli range can reproduce much more of the subtleties of playing an instrument than weā€™ve had before. Physical modelling and the expressiveness of MPE are only going to get better over time. While modelling will never ā€˜replaceā€™ physical instruments, over the coming years weā€™ll be able to approximate them ever more closely.

2 Likes

Mostly cause I canā€™t afford a selection of 40 pianos and various halls to record in, sample libraries more than make the cut for me though

2 Likes

If I was a keyboard or piano player thereā€™s no way Iā€™d be satisfied with samplesā€¦ but in the end it depends on the genre. If youā€™re making commercial music youā€™re unlikely to care - and 90% of your audience certainly wonā€™t.

I see the MPE controllers not as a way to emulate real instruments (a synth will never be a violin) but as a way to play synths with a lot more expression than traditional keyboards, to make music that has never been heard before.

1 Like

Because the production of commercial music for mass consumption requires cutting out the expenses associated with using real musicians, because corporations only care about profits.

I want more PM hardware synths, basically, but I donā€™t care if they sound like a real saxophone player or not. Itā€™s when you can push and bend and break the barriers at the edges of Physical Modeling that it becomes fun. (Kindof like how the old Vector synths were more fun than just their ROMpler sounds since you could find ways to creatively layer those and some purely synthesized sounds).

I guess what I really want is metaphysical modeling. :slight_smile:

7 Likes