I think using the SOTA model of 11labs actually solved my problem. Thanks @Mason ! Also what would you recommend as an alternative for 11labs for good quality voices?