festvox-kdlpc16k binary package in Ubuntu Focal arm64

 This is a diphone database for festival that uses 16k samples for high
 quality sound output. This is a large database (9 MB), and festival will
 convert text to speech more slowly using it, so you might want to install
 festvox-kdlpc8k instead if you have low disk space or a slow computer.
 .
 The voice contained in this database is an American English male speaker.
 A different American English voice can be found in the festvox-kallpc16k
 package if you dislike this one.

Source package