AI Voice Modulation in The Witcher 3
The Witcher 3 is receiving a new expansion more than 11 years after its initial release. To accommodate health challenges faced by Geralt's Polish voice actor, CD Projekt Red is employing AI technology to modify his vocal performance.
Jacek Rozenek, the Polish voice of Geralt, is returning for the Songs of the Past expansion. Following a stroke in 2019 that impacted his voice, the studio is using the Respeecher AI tool to align his current vocal output with the character's 2014 performance.
Rozenek previously returned for quests in the 2022 Next-Gen update but noted significant exhaustion during the recording process. To facilitate his return for this new content, the studio sought alternative methods to support the actor.
Maintaining Vocal Consistency
The primary challenge was the discrepancy between Rozenek's current vocal range and his 2014 recordings. His voice is now lower in pitch, which would create inconsistencies when placed alongside older dialogue. Respeecher was implemented to bridge this gap.
Synthetic speech artist Dmytro Vasylets explained that the process involved dozens of hours of model training and meticulous work to address AI artifacts, moving beyond simple filtering.
Actor Consent and Industry Context
Rozenek provided his consent for the use of the tool, expressing continued trust in the development team and satisfaction with the acting dynamics. CD Projekt Red emphasized that this implementation is a specific, consensual use case and stated they would not use such tools without an actor's express permission.
The studio clarified that this technology will not be required for The Witcher 4, as the narrative will feature an older, more grizzled version of Geralt where a deeper voice is appropriate.




