It has not been very long since my last entry, but I thought it was time to write one anyway because I have just managed to get semi-voiced sounds working. Technically unvoiced sounds work as well but they are currently way too loud so I'm ignoring them for now.
First, as a reminder, here is what the phrase I was using to test yesterday sounded like at the normal speed:
And now, it sounds like this:
We still have the issue with the vowel in the word visual, but that is expected since I haven't done anything to try to fix that yet. However, the most significant difference here is the introduction of the noise signal at appropriate points in the audio. At least to my ears it now sounds quite a bit clearer and I don't have to pay quite as much attention to understand the speech.
You may also notice that the beginning and end of the phrase are smoother as well. Since I had to add envelopes in order to transition between voiced, semi-voiced and unvoiced sounds anyway, I took the opportunity to add a short fade in at the beginning of each phrase and a fade out at the end. Currently it is set to 35 milliseconds or half the duration of the phone (whichever happens to be shorter). I might have to have separate settings if the first and/or last sound is unvoiced, but that's a problem for another day.
As always, we will conclude by listening to the same phrase again but at the slower speed:
I plan to improve some of the semi-voiced filter settings a bit as I'm not 100% happy with them, but for now I will pat myself cheerfully on the back and get some well needed rest.