CD Projekt Red Uses AI to Restore Voice of Geralt's Polish Actor Following Stroke

When CD Projekt Red began localizing The Witcher 3: Songs of the Past for its 2027 release, the studio faced a critical casting hurdle. Jacek Rozenek, the Polish voice of protagonist Geralt of Rivia, could no longer deliver the character's signature gravelly tone.[1][2]

Rozenek suffered a severe stroke in May 2019 that left him with aphasia and fundamentally altered his vocal range. Rather than recasting the iconic role for the upcoming expansion, the development team opted for a technological intervention.[2][3]

The studio partnered with Respeecher, an artificial intelligence voice synthesis company, to process Rozenek's new studio recordings. The collaboration allows the actor to continue performing the character he has voiced for nearly two decades.[1][4]

Rozenek remains fully active in the recording booth, acting out the localized script and providing the emotional foundation for the performance. The AI software then acts as a sophisticated post-production filter.[1][5]

A bespoke technical solution

The Respeecher model was trained on hundreds of hours of Rozenek’s pre-2019 audio files from previous games in the franchise. It modifies his current delivery to match the exact timbre and pitch of his original Geralt performance.[1][4]

CD Projekt Red emphasized that the actor was heavily involved in the decision and gave explicit consent for the voice synthesis. "This was a special case," the studio noted, distinguishing the project from automated generation.[2][5]

The actor's recovery from the 2019 stroke has been a long process, initially involving right-sided paralysis and extensive speech therapy. Retaining his role in Poland's most famous cultural export provides a significant professional milestone.[2][3]

AI

The gaming industry is currently locked in a fierce, high-stakes debate over the use of generative AI and synthetic media. Actors and unions have repeatedly clashed with major publishers over the threat of automated replacement.[1][2]

Navigating the AI debate

Recent SAG-AFTRA strikes and ongoing contract negotiations have centered heavily on studios using synthetic voices to cut labor costs. Voice actors fear that AI models trained on their past work will eventually render them obsolete.[1][4]

CD Projekt Red’s approach flips that narrative, deploying the technology as an accessibility tool rather than a mechanism to bypass human talent. Rozenek is compensated exactly as he would be for a standard, unassisted recording session.[2][5]

The Polish developer has navigated this specific ethical territory before. In 2023, the studio used Respeecher for the Cyberpunk 2077: Phantom Liberty expansion to recreate the voice of Miłogost Reczek, a Polish actor who passed away in 2021.[1][2]

In that instance, the studio secured explicit permission from Reczek’s family and hired another actor to perform the lines before applying the AI filter. The family's approval established an internal precedent for ethical AI deployment.[2][4]

The 2027 expansion stakes

Songs of the Past represents a massive undertaking for the developer, arriving a full 12 years after the base game launched. Maintaining continuity for long-time fans of the franchise was a major priority for the audio team.[1][3]

Rozenek has voiced the monster hunter in every Polish-language release since the original The Witcher debuted on PC in 2007. His performance is deeply ingrained in the local cultural footprint of the series.[2][3]

Recasting the role would have been a jarring shift for the domestic audience, making the AI restoration a highly practical solution. The technology bridges the gap between the actor's current physical capabilities and the character's established identity.[3][4]

AI

Respeecher has built a strong industry reputation for high-profile, actor-approved voice restoration work over the last five years. The firm previously collaborated with Lucasfilm to synthesize voices for legacy characters in various streaming series.[1][4]

Setting an industry standard

Voice actors and labor advocates are watching the CD Projekt Red collaboration closely as a potential model for future union contracts. It demonstrates that AI synthesis can actively extend a performer's career following illness or injury.[2][5]

The crucial distinction for labor groups is that the actor retains ultimate control over their digital likeness and receives full financial compensation. The AI serves the performer, rather than the studio circumventing them.[4][5]

As the 2027 release window approaches, the audio engineering team is finalizing the localized tracks across 15 different regional markets. The Polish localization remains a focal point due to the franchise's origins.[1][3]

The success of Rozenek's vocal restoration could dictate how other major video game publishers handle aging or injured legacy cast members. The project provides a rare, concrete example of artificial intelligence preserving human employment.[2][5]

Finalizing the audio mix

The technical pipeline requires meticulous audio engineering to ensure the synthesized voice blends seamlessly with the game's dynamic environmental acoustics. Sound designers must match the AI output to the spatial reverberations of digital taverns, caves, and open forests.[1][3]

This level of audio fidelity is essential for an expansion expected to demand dozens of hours of player engagement. Any artificial artifacts or robotic clipping in the dialogue would immediately break the immersive experience of the fantasy world.[3][4]

By prioritizing Rozenek's continued involvement, the studio has turned a potential production crisis into a testament to technological accessibility. The final audio mix will stand as a unique intersection of human resilience and advanced software engineering.[2][5]