Hi! First of all, apologies if this question is too strange or off topic... I have been using Speex as a vocoder, with narrow-band mode and quality 0. I think the results are very interesting, especially when feeding with non-speech data. I tried a little to find some parameters to tweak, to make it sound more like a TI-99 or a Speak & Spell (maybe especially to make the pitch prediction worse), but found it difficult - when I change something, either things sounds as before, or are completely broken. I wonder if any of you perhaps ran into some interesting accidents while developing, that I could try to reproduce? Or some good constant to tweak? All the best, Johan --- ?u????u??u? ???u??? ?n??u??u?-?o?d?p ??su???o? u??o?