Open Zireael07 opened 10 months ago
In theory it could, but you would need the formant frequencies for the additional phonemes (F1, F2, F3, with F1 and 2 being the most important generally.)
There would also need to be durational data for each phoneme and if the language is tonal (like English and Chinese are) you would also need data for F0 fluctuation.
Then to really make it intelligible, you need cadence (variation in duration and pitch at the sentence level) data. The data for Speako-8 was taken from this book: https://archive.org/details/frontiersofspeec0000unse
I like the way it sounds! Can this be extended to support non-English phonemes? Where's the phonemes data?