Fungisonoris wrote:I’m a amateur musician who for many years have been developing a design for a software synth, mostly out from my own preferences and wishes. Now I’m getting together a team to realize it. I have come to the conclusion that to have a good chance on the market there needs to be a demand for the concept and features. Maybe there would be if they were known, but I don’t want to impose any bias to the question.
I would like to know if there are things you miss or could be better among the synths today. Is there any concept or features you think should be out there but is missing?
Maybe this question is improper in this forum? I don’t know. Any suggestions where I else should turn with a question like this?
You fall in a cave of trolls...
It's unlikely you will get any representative answers (which is something you probably already guessed by now).
That said (and seriously) build your synth around samples and sample manipulation. Give it like four oscillators, each oscillator would have either a single sample (which can act as a waveform or not, depending on the way you configure the oscillator) or multi-sample maps.
You could enrich it with unison features, although I don't think this is really necessary, given the high complexity the synth is already capable of because being sample based.
The oscillator could also be able to analyze and resynthesize the sample, creating an additive representation, using either the transient sounds or the sustained sounds, with just the noisy components or the harmonic components or both (the analysis/resynthesis portion of SoundHack is a good example of what I am talking about). It could do granular synthesis too (based on samples or on sine tones). And it could create a wavetable too (given the resynthesis capabilities, this is somehow implicit).
The filter part should have composite filters, like comb filters, spectral filters, formant filters, modeled resonant instrument filters, etc. Not just the old and tired low pass, high pass, bandpass - enough of that. Each oscillator should have it's own complete synth channel, with filter and amplification, basically working as a complete layer before the effects, in the Roland way, although we could also route the oscillators to single layer, or two to one layer and two to another layer (basically we should have a configuration page to define the routing - and since we have a configuration page, we could also implement some other oscillator interactions).
Of course, the interpolation and transposition algorithms had to be the best, because I would want the sound to be top notch across the entire spectrum, even if just using a single sample.
Plenty of modulation sources (ADSR envelopes, multi-segment envelopes, LFOs, sequencers, etc) and a modulation matrix. Many oscillator functions should be able to be modulated.
A good creative bunch of effects (Falcon is a good example on this) to complete the cake. And that's pretty much it.
Oh, and NO FLAT interface. Invest in a good designer to create a good interface, with soft colors (no red over black... actually NO RED at all) which is as clear as possible and skeuomorphic enough to give us an idea of depthness - I simply HATE flat interfaces.
If you follow these ideas, you would create a hell of a synth... and you will probably sell just a few, because the vast majority would not be interested in such a beast (they are more interested in emulations of Minimoogs and TB-303s).
I would, though.