Igro wrote:Also, I would love to see more wavetables added. The reason for this is the high quality of the factory wavatables. When I import mine, the result is not so good.
You have to think about what all the technicalities mean: What is the meaning of a additive wavetable slot with the resolution of 256 partials? So the highest partial/frequency needs at least 2 consecutive samples, i.e. one for the peak of a sine and one for the minimum of a sine. Could it be that if you want to sample from one of your favorite synths, the sample should be sampled from an F#4. Not exactly sure about the 4 in F#4, but in general most wavetable synths over a lenght of 128, 256, 512, 1024,... samples in lenght. If you compute it this means a F#n note...(A full cycle sine of 2048 sample does fit how often into 44100 samples per second? Then look up frequency in a frequency to note table). This is just to give you some sort of intuition about how things work in sample based wavetable synths like Tone2 Icarus.
Thorn is only "similar" but not exactly "the same" because of it's additive oscillators.
Here I describe how I made my speech patches for Thorns additive oscillators.
The process works like this - all what I say is more or less intuition based
1.) Recorde the source sample from balabolka, a text-2-speech programm
2.) load it into audacity, trim it
3.) Check out the lenght in samples, let's say it's n-samples in length.
4.) I want to feed each slot of Thorn with only 512 samples, so I compute 512 * 16 -> Target Size. 512 sample is a compromise, i.e. you need to find a good value that gives you a good resolution in terms of time and frequency.
5.) I compute a resampling rate, i.e. applying the resample -rate to the n-samples will give me 512*16 samples. Note: Applying the resampling means you lose quality if your n is larger than 512*16! The larger the difference is the more quality you lose.
6.) The resampled sample now with length 512*16 I split into 16 regions in audacity (or was it wavosaur?)
7.) Exported all regions into a wav file each
8.) Import each region into one slot one by one
9.) Setup a envelope that scans thru the slots linear fashion, so recreating the original sample
This works best, if the sample you start with is not to big, or other way round as close as possible to 16*512 samples...Thus you might want to start with the right sampling note...
BTW: you can do this speech sample approach with any synth that offers oscillators with wavetables which over slots. I have done a similar thing to U-HE Zebralette (check out: https://soundcloud.com/p-e-t-e-r-h/osc- ... kas-attack
Some very good synths even offer a dedicated "import sample / resample" function, thus you do not have to do it on your own, like I did. Check out Tone2 https://www.tone2.com/icarus.html
or I think Avenger and Parawave Rapid have similar import functions.
] Peter:H [