|
|
sáb 02-abr-2005 2:08
Espero que lo entiendan, no me da para traducirlo todo - habla de los soft synths y de por qué a veces suenan feo.
--------------
It is interesting that there is always a latency argument in software verses hardware : hardware can have a lot of latency in it.
Take for example one of my favorite hardware synths/samplers the ensoniq mirage. That has a single processor in it : one of the jobs it has to do (as well as playing back samples) is to scan the keyboard. If you hit 6 or seven keys down at one when playing a chord or similar, it dosent have enough processor cycles to scan the keybaord register, and play the notes, and so the led display disapears.
To suggest any keyboard scanning synth would have a sub 5-10 ms latency I think would be incorrect.
The same goes for midi. PLay a module from a controller keyboard, and you get a good few milliseconds of latency via midi as it encoded and then decoded at the other end.
Now it is possible deal with latency, everyone is paranoid !
I dont know off the top of my head what the latency between hitting a piano key down and the start of the note sounding, but it isnt zero.
As for how a soft synth sounds : it depends which ones.
I like applications like reason and reaktor, but I dont like VSTi's because they are non-modular and cant interact with each other.
Most soft synths sound great in isolation, but making a full mix, things start to fall apart, not because the insruments sound bad, but becasue they allow to jump a few steps ahead of where your head should be at.
(sigue)
|
|
|
sáb 02-abr-2005 2:09
1) VSTi, DX, AUi's cant interact. Each happen in their own sonic space, and dont tend to bind well. Also the way polyphony is normally generated, by cloning the proccesses into more memory space means that the indiviual notes cant react to each other either.
2) Phase issues : There arnt any. Everything is phase locked from lfo's to ocillator ramp starts, to EG curves. This is very unnatural and your ear pulls the mix apart as it is very sensative to phase.
3) The tuning knifeedge : everything is 100 in tune and locked to even tempered scale by default. This tuning was never designed to be anything other than a compromise, and when you have everything tuned to 100% accuracy, the musical scale itself starts to show its weaknesses and start to sound ugly.
4) The grid of doom : not you dont have midi timing drift, and you are 100% phase locked, you can quatize something (like drums) so it it is not only 100% locked to the grid, but in the case of the drums, the envelopes are phase locked against each other in the layering. Quantize the peaks too, and you will end up with all the attack traniants perfectly alligned in time and phase and tuning.
So now you cant here the attack transients, and your ear can no longer recognise the quality of the individual sounds. And watch that master level peak on the down and up beats.
And I havent even started on effects......
But by simply /not/ doing the above, you can make something as good as the best hardware, delaying notes, tracks, offsetting the tuning on tracks etc adding randomness back in (or even better "groove")
Most of the software only music I hear suffers from the basic faults i have described above, and is giving soft synths a bad name, which is not because software instruments sound bad, but because they people using them dont know enough, or think enough about acoustics and physics.
Using VSTi' in render mode feels like the right thing to to do, but you miss the fundimental principle in synthesis whereby what you are trying to do is move air in a pleasent way.
And if you look for some research on how people percieve sound as being natural or organic, softsynths rendered directly to disk by its nature will natively produce a mix which you brain will pull apart as being wholey synthetic.
Sometimes this is what you want, but most people will feel uncomfortable, or unengaged with music that shows these traits.
|
|
|
sáb 02-abr-2005 2:10
Lo saqué de un foro en inglés, claro
El consejo que daban: en lugar de hacer render grabar los softsynths en tiempo real por la entrada de audio...
Sólo hacer render cuando tenés los tracks grabados de hardware.
Me parece interesante.
|
