is useful for zynthian or just cost too much?
Depends on exactly what you want to do really. Apart from loading in the lunatic 4GiB linuxsampler sound fonts I can do pretty much everything I want with a 1GiB Pi4. I’d upgrade to a Pi5 for Surge XT if I felt it was compelling, but that’s only for more processor cycles.
With the Raspberry P5 8Gb I was able to max out the PianoteQ settings which now has a completely different and better sound than the P4 8Gb… for me who uses PianoteQ 99% compared to other sounds, it is essential. Maybe 16 Gb is really a lot… but technology is moving fast like a train. The thing that makes me happy is the fact that we are moving to wonderful sounds that no longer rely on samples and are therefore much lighter.
Personal opinion:
Unless you plan to use the zynthian live with huge sampling libraries or as a full mixing and mastering DAW, you will hardly benefit from 16 GB, while spending quite a bit.
A 4 GB should be already enough, the 8 GB is a solid and future proof choice (you should be able to get the latter new for 60-70€ with shipping included from ebay).
Consider that with about 140€ you can get the 8GB RPI 8, a KKSB / inch display case and a 7 inch official or compatible touchscreen.
The thing with technology is to buy what you need, when you need it. Buying over spec is rarely economically beneficial because when you need the extra resources in the future, they are much cheaper.
I posted the video from Jeff to give us a look on his opinion about 16gb pi. He made some great points. Unless zynthian adds something that needs 16gb ram I don’t think it’s needed but if you plan on doing a lot more with zynthian and see a need for 16gb than by all means get it.
Ollama AI 100 pc orchestra auto accompianiment.
You maxed out pianoteq with 8 GB? Better sound?
what did you do?
Perhaps you can upload a maxed out example so I can load it into piaboteq?
Cheers and thanks,
Maarten
For my taste now the sound is more … crystalline, more … explosive on the bass, without muddying. But this is my opinion that depends on the person. Obviously I’m using Raspberry P5.
To put it to the maximum on Raspy on the WebConf page put these options. Bye
I wonder if @Lanfranco means “maxed out” to be Pianoteq configured to use its full potential whilst @maartmaart may consider “maxed out” as meaning Pianoteq has exhausted its available resources?
With raspberry P4 8Gb, pianoteQ set to maximum from WebConf, it did xrun. With Raspberry P5 8Gb it works fine and the sound, to my ears, has improved.
Hi @Lanfranco and @riban,
I was wondering how more memory (with the some piano configuration settings) can influence the sound quality of pianoteq. I can imagine more notes and more fluent user intercae etc… but sound quality?
Cheers,
Maarten
ow… a different PI…
Then I still do not understand… except for more sound resonance due to more notes…
Sorry… should learn to read…
Cheers,
Maarten
…and I should learn English… . maybe it’s my impression, but the sound seems richer in harmonics… In practice, when I use PianoteQ with the iMac in my Home recording room, it has a very crystalline sound… while with the raspberry P4, it was more closed. Now, with the raspberry P 5 and all the functions set to maximum, I no longer hear the difference between Zynthian and iMac… but my old ears could also be wrong…
You don’t need the extra memory. A RPi5 can run Pianoteq at full audio bandwidth and full polyphony. Older models had to limit the internal sample rate, halving the audio bandwidth, e.g. 48000 frames per second was set to 24000 fps internally, resulting in hf roll-off at 12khz. This makes the sound dull.
That explains it… lower sampling rate… results in less defined sound…
Thanks @riban for clarifying this…
Hi everybody
My 2 cents on Pianoteq.
It definitely runs without a glitch on Pi5 at 48.000 Hz of sound generation cycle. I don’t think that 16 Gb would make any difference for Pianoteq, since 4 Gb is already enough for running comfortably the physical modeling algorithm. More RAM comes handy if one runs Pianoteq and a certain amount of heavy soundfont libraries, for example to set up a template for classical/cinematic music, which, if remaining within Zynthianland exclusively, would probably require two devices (or more, depending on orchestral complexity and variety of articulations).
The problem with Pianoteq at half sample resolution is that, with anti-aliasing filters kicking in beyond 12 Khz, the strongest highlight of the engine is lost, that is its virtually endless sympathetic resonance between mathematically modelled piano strings, which tend to have high frequency ringing sonic spectra. Hence, the double effect of muffled/opaque sonic result, both in frequency resolution and loss of sympathetic resonance.
As a pianoforte instrument in itself, I don’t think that Pianoteq is right for everybody and for every genre of music. There is a certain crystalline and punchy character to its overall output, that lends itself quite nicely to mainstream jazz, rock and pop, but not so much to classical writing or more pensive music (like, say, ECM Scandinavian stuff in general). In my opinion and by direct experience, it sounds quite convincing in the context of a larger arrangement, but not as much as a solo instrument, especially when slow moving harmonies with a prominence of lows and mids are concerned. To my ears, its central range sounds rather artificial (which could even be a subjective strong point) and lacks a bit of the feel, weight and gentleness of wood, more notably in the dynamic mp-mf territory, while it absolutely excels at metallic ringing detail, precise shimmering arpeggios and edgy strong basses.
I have had the chance of playing digital pianos with hybrid engines (physical modelling manipulating samples) that, by any objective standards, stand head and shoulders over Pianoteq, but as a technological achievement per se is a resounding programming and engineering success, because it gives us the luxury of a pretty realistic portable piano, delivered by a lightweight software that runs smoothly on average computer hardware.
Kind regards
As an amateur, I can say that my piano experience began with a rather scandalous upright piano… then I moved on to the legendary Yamaha CP 70b, which had ONLY the keyboard as a good thing, but the sound… in my opinion it was ugly (but I played a lot of concerts with my Italian rock band). Then some Yamaha digital pianos, Roland… For me, the digital piano must respond to the dynamics and make you feel that the sound changes from pp to ff, while the digital pianos I had, only increased the volume of the sound. for this reason I stopped playing the piano and took up the trumpet… for me, the sound and dynamics are very important, despite my old ears. I discovered PianoteQ thanks to Zynthian.
I think modeling instruments are the future (apart from the wonderful analog synths)… just try the SWAM winds to understand it…
I don’t think sampled sounds like Native Instruments’ Symphonic Orchestra would be able to be used on anything less than very powerful machines.
I concur @Lanfranco, they are, but they will also require a further technological leap, both in hardware specs and programming development (like contextual predictive AI, capable of inferring timbral and spectral parameters from the Midi flow), which I think is still quite far from the current state of things. SWAM instruments are nice by the way, but still miles away from the subtleties and dynamic nuances of their real acoustic counterparts, at least in my opinion.
Regards