Microsoft VALL-E can generate realistic imitations of human speech from samples as short as 3 seconds January 10, 2023 // by Finnovate This content is for members only. Sign up for access to the latest trends and innovations in fintech. View subscription plans. Login