3 ms·

Physically modeling a Rhodes somewhat accurately is beyond the computing capabilities of your typical laptop computer, so such models are always highly reductio

Physically modeling a Rhodes somewhat accurately is beyond the computing capabilities of your typical laptop computer, so such models are always highly reductionist, deliberately so. Source: I work with 2D meshes to models and even a single physical object can get almost infinitely deep and complex if you want to make it so. A Rhodes has a set of metal tines that get plonked like a ruler on a table edge, rather like the tines on music boxes, and there’s a set of two tines, each with its own pickup for EACH NOTE on the keyboard! https://www.fenderrhodes.com/service/tine-settings.html https://www.fenderrhodes.com/service/tine-settings.html https://www.vintagevibe.com/blogs/news/history-of-rhodes-pickups https://www.vintagevibe.com/blogs/news/history-of-rhodes-pic... Of course, you can use a sampled Rhodes and it’s only noticeable to either the player or if it’s not buried in a mix of a zillion other tracks, there is a notable lack of depth to it. Like most things on computers: “good enough” wins the day. I switched to purely virtual synthesis in 1999, and have spent the last 5 years revisiting physical musical instruments as I lament the electronification of music I was such a fan of for 20-30 years

ushakov 4y ago
Timbre-Transfer with Machine Learning (see: magenta) will allow to play exact digital copies of instruments on your typical laptop computer Keith Bloemer of GuitarML has already done this with guitar amps (https://github.com/guitarml https://github.com/guitarml)
jalk 4y ago
So ie pianoteq models are very cpu intensive and/or not very “accurate”? (genuine question)
- netr0ute 4y ago
  Pianoteq now isn't meaningfully CPU intensive because our CPUs have gotten so fast (read: Alder Lake and Zen 3)
  - dietrichepp 4y ago
    I routinely hit 100% of a core on Pianoteq on a 3 GHz i5 from a couple years ago (keep in mind it uses multiple cores). Pianoteq is definitely CPU-intensive. Diva hits around 50% and it's also considered CPU-intensive.
    - netr0ute 4y ago
      > 3 GHz i5 from a couple years ago That's nothing compared to what's readily available today; this isn't 8 years ago when CPUs were stagnant.
      - dietrichepp 4y ago
        If your assumption is that people are always buying brand new Threadrippers or Mac M1s every couple years, then we're living in different worlds. Musicians, especially, are much slower to upgrade hardware and software than other users because of the risk that something breaks during the upgrade--there are just so many different pieces of software that can break. A lot of musicians with Macs stayed on Mojave for a few years after it became obsolete just so they could run 32-bit programs, even though the writing has been on the wall for 32-bit programs since the very beginning of the Mac on x86 era.
        netr0ute 4y ago
        I'm not one of those who uses software that can't just be recompiled or emulated for a newer platform.
prmoustache 4y ago
The fact there are differences between a software and hardware synth doesn't mean that: - you could spot it isn't the real one outside of a comparison test. - you would prefer the real one in a blind hearing test - it matters once the sound has past through effect, the composition has been mastered and compressed. - it matters once lossly streamed, playing on the typical shitty bluetooth speakers and headphones while in a noisy environment which accounts for the majority of the context where records are played.
- maldusiecle 4y ago
  Surely a serious musician is going to want to create music that rewards attention, not music expressly built for distracted people on shitty speakers they can barely hear.
  - musicale 4y ago
    > Surely a serious musician is going to want to create music that rewards attention Not always. Consider Robert Rich's very serious idea of "egoless" music which isn't intended to command attention but to leave space for the listener's own internal attention and creation. In his sleep concerts listeners even fall in and out of unconsciousness, presumably experiencing some interesting dreams as well. > music expressly built for distracted people on shitty speakers they can barely hear An apt description for many things, from wax cylinders to AM radio to MP3 to to music played on smartphones. Which is to say: all popular music since the 1890s, "serious" or not.
- munificent 4y ago
  You're evaluating an instrument purely from the perspective of the listener and not the musician. Sure, an emulated Rhodes might be indistinguishable once it's gone through the audio production and mastering meat grinder. But the real question is whether a musician will even want to put one through that if the sound doesn't inspire them the moment they sit down and play it. I use a mixture of analog and digital synths. While I appreciate the power and flexibility of emulations, I've never been inspired to make music as much as I am when I press a key on an analog synth.
PaulDavisThe1st 4y ago
> Physically modeling a Rhodes somewhat accurately is beyond the computing capabilities of your typical laptop computer, so such models are always highly reductionist, deliberately so. This really not true anymore. Pianoteq includes several e-piano models and it runs just fine on a typical laptop. It also includes physically modelled acoustic pianos, harpsichords and several other instruments. All run fine on a typical laptop.

ushakov 4y ago

Timbre-Transfer with Machine Learning (see: magenta) will allow to play exact digital copies of instruments on your typical laptop computer Keith Bloemer of GuitarML has already done this with guitar amps (https://github.com/guitarml https://github.com/guitarml)

jalk 4y ago

So ie pianoteq models are very cpu intensive and/or not very “accurate”? (genuine question)

netr0ute 4y ago
Pianoteq now isn't meaningfully CPU intensive because our CPUs have gotten so fast (read: Alder Lake and Zen 3)
- dietrichepp 4y ago
  I routinely hit 100% of a core on Pianoteq on a 3 GHz i5 from a couple years ago (keep in mind it uses multiple cores). Pianoteq is definitely CPU-intensive. Diva hits around 50% and it's also considered CPU-intensive.
  - netr0ute 4y ago
    > 3 GHz i5 from a couple years ago That's nothing compared to what's readily available today; this isn't 8 years ago when CPUs were stagnant.
    - dietrichepp 4y ago
      If your assumption is that people are always buying brand new Threadrippers or Mac M1s every couple years, then we're living in different worlds. Musicians, especially, are much slower to upgrade hardware and software than other users because of the risk that something breaks during the upgrade--there are just so many different pieces of software that can break. A lot of musicians with Macs stayed on Mojave for a few years after it became obsolete just so they could run 32-bit programs, even though the writing has been on the wall for 32-bit programs since the very beginning of the Mac on x86 era.
      - netr0ute 4y ago
        I'm not one of those who uses software that can't just be recompiled or emulated for a newer platform.

prmoustache 4y ago

The fact there are differences between a software and hardware synth doesn't mean that: - you could spot it isn't the real one outside of a comparison test. - you would prefer the real one in a blind hearing test - it matters once the sound has past through effect, the composition has been mastered and compressed. - it matters once lossly streamed, playing on the typical shitty bluetooth speakers and headphones while in a noisy environment which accounts for the majority of the context where records are played.

maldusiecle 4y ago
Surely a serious musician is going to want to create music that rewards attention, not music expressly built for distracted people on shitty speakers they can barely hear.
- musicale 4y ago
  > Surely a serious musician is going to want to create music that rewards attention Not always. Consider Robert Rich's very serious idea of "egoless" music which isn't intended to command attention but to leave space for the listener's own internal attention and creation. In his sleep concerts listeners even fall in and out of unconsciousness, presumably experiencing some interesting dreams as well. > music expressly built for distracted people on shitty speakers they can barely hear An apt description for many things, from wax cylinders to AM radio to MP3 to to music played on smartphones. Which is to say: all popular music since the 1890s, "serious" or not.
munificent 4y ago
You're evaluating an instrument purely from the perspective of the listener and not the musician. Sure, an emulated Rhodes might be indistinguishable once it's gone through the audio production and mastering meat grinder. But the real question is whether a musician will even want to put one through that if the sound doesn't inspire them the moment they sit down and play it. I use a mixture of analog and digital synths. While I appreciate the power and flexibility of emulations, I've never been inspired to make music as much as I am when I press a key on an analog synth.

PaulDavisThe1st 4y ago

> Physically modeling a Rhodes somewhat accurately is beyond the computing capabilities of your typical laptop computer, so such models are always highly reductionist, deliberately so. This really not true anymore. Pianoteq includes several e-piano models and it runs just fine on a typical laptop. It also includes physically modelled acoustic pianos, harpsichords and several other instruments. All run fine on a typical laptop.