“Anatomy of Flex Fm” section under “sampling”
Well, the manual says (I quote literally, so you should be able to copy&search to find the exact spots):
[Sampling] 8 mono voice sample pools at 44.1 kHz, about 13 seconds each, with up to 16 slice markers (8 by default)
Samples and kit pads share the voice’s roughly 13-second pool—not a separate pool for each operator or pad.
I might have missed other mentions (and possible limitations - though I think e.g. HIFI engine or STEM mode (workflow?) does not limit the sample pool length, but the number of sampling operators you can use at the same time within a voice), but above should be the gist of it.
It certainly sounds short - but then again, there seems to be a lot you can do with the (short) samples to keep things interesting, and limitations foster creativity 
1 Like
If you add the manual to https://notebook.google.com/ you can easily search or ask for information there. It’s really good and I highly recommend 
1 Like
Thanks. It’s a really really low time for sampling.
I’ve been reading a little from the manual but it seems too cryptic for me.
As I understand each voice (or track) has 4 operators and each operators can be a synth, a sample, or maybe chords with various operators working at the same time. Is this how it works…
I’m still recovering for the surprise of the low little time for samples… It’s too short but let’s see the next videos…
I hope it will be true. I don’t mind sacrifice some voices to be able to load more samples to the pool.
13 seconds per voice seems fine to me, tho of course ymmv. a drum break would be the longest sample I’d be using per track.
1 Like
A drum kit with lots of hats/cymbals/crashes with long releases might be difficult - but then again, creativity would have to take over, e.g. splitting such kits over multiple voices etc. (or synthesizing a long release with other operators and/or sample playback loops).
I am currently not worried about the 13 second limitation (if I need to play longer samples, I probably want a different machine).
In total it seems you have 8 voices x 13 second sample-pools = 104 seconds of sample time per song (i.e. project - thanks @pidgeo for the clarification) pattern (a little simplified - I think it’s actually a little more complicated). And on top of that (very simplified) 3 other operators per voice for “synth” stuff.
Don’t know if that is enough, but seems plenty to me.
3 Likes
You’ve explained so well. Thanks!
that should be per song afaik. patterns in flexFm are just sequence data contained in a song.
1 Like
Yes, you are right, thanks.
SONGholds the complete project: its title, eight voice patches and samples, patterns, chain, mix, effects, tempo, scale, and key
I’ll update my previous post to avoid misunderstandings.
1 Like
Yep that’s what I understand too.
A pool of 13 seconds per voice and a voice is 4 op , each operator can play a sample unless you are using HI FI mode , 1 op only. So you have 8 x 13 seconds in total.
That’s enough for me , I hate waiting for samples to load in ram.
A voice can also play a stem for long audio ( streaming fromHD)
I think some of the limitations about nb of patterns , samples etc come from the fact that you can morph sounds , patterns etc.
2 Likes
I suppose although it contains 32GB internal storage, it depends on how much RAM the device has? You could have 1TB internal storage but if the RAM is tiny, then samples used per project is still going to be very limited, as there is not much can be done unless sample streaming from the internal storage is implemented.
The Polyend Play has this issue, it uses a MIcroSD for sample storage, but only around 30mb sample pool per project, which is tiny.
I think the DT2 hit’s the right mark with a decent 400mb per project.
ffs what’s all this whining about samples?
buy a sampler if you’re after that. it’s an FM groovebox featuring the most flexible implementation to be available on the market. and again it’s not even released and no proper audio demos been heard, but the requesting of features “it should have” and that are for sure “easy to implement” goes wild. why not just wait and see? the developer(s) for sure had a very respectable vision and reasoning to design it the way it will be. if you ask me, it should be treated exactly like that.
until then: do some music.
peace. 
17 Likes
Looking forward to hear what will wizards(witches) like @Jeanne do with it. (ofc if she’s buying one)
2 Likes
It is definitely the most tempting thing I’ve seen in ages. Features I like:
FM with virtual analog waves
4 LFOs, drawable
Saw Stack
Samples combined with Oscillators
Free tempo per Track (e.g. Polyrhythms as Quintuplets)
Endless Encoders
Delay Effect has feedback with resonance
Reverb Effect has Shimmer that can be tuned to something else than (ewww) octaves
Lots of Filter Types (though I wish choosing a digital filter wouldn‘t replace the analog filter)
Filter Morphing
ADSR Envelope with control over the shapes
Joystick
Piezo velocity sensitive drum pad
6 Likes
all great stuff 
I’ll add -
voice freezing/render to sample.
resampling.
slicing
I like them ones too
I admit I am also of the camp that I wish it had more sampling. Mostly it comes from this being so close to perfection. Of course I know that you can go far with 13 seconds of sample time. I have a little device that is 20 seconds total, and you can do a lot with it. So ultimately I think the limitation is fine. We are all just a greedy bunch.
1 Like
would love if some kinda stretch algorithm was possible in the future. like this MD custom user machine wan. would be killer.
(Just for the record, I‘m in the camp of anything sampling adds 0% to my excitement 
)
3 Likes