ARC Raiders has replaced more AI-generated dialogue with human performances, but Embark Studios has not confirmed that every synthetic voice is gone. The developer says it has re-recorded its quest lines with actors wherever possible. It still defends text-to-speech (TTS) as a production tool, particularly for the game’s constantly changing ping system.
For players who thought ARC Raiders was moving away from AI voice acting, that explanation is frustrating. After months of hearing about new recordings and better character performances, why is Embark still making the case for the technology it has been replacing?
AI Voice Status
As of November 4, the clearest confirmed change concerns quests. Embark says it has re-recorded all quest lines with as much of its acting cast as possible. Earlier updates had already replaced some generated dialogue, and the studio acknowledges that human performances sound better in important scenes.
In a November 3 interview with The Game Business, Embark CEO Patrick Söderlund went further. He acknowledged that some text-to-speech lines had received complaints about their poor quality. The studio decided to re-record those lines with real actors, and Söderlund said the criticized lines are now gone from the game.
However, his comments did not clearly establish whether every AI-generated line has been removed, especially from the dynamic ping system.
At launch, ARC Raiders used a hybrid approach: most dialogue came from voice actors, while certain quests and mission briefings used AI-generated speech. Contextual pings combined recorded fragments with lines generated by a model trained on actors’ voices.
Embark still describes TTS as useful for those pings. However, it has not released a current list identifying which live callouts, NPC interactions or other lines are synthetic. Its defense of the system does not prove that every ping heard today uses AI. Equally, the quest re-recordings do not establish that AI speech has disappeared from the game.
The Renewed AI Voice Backlash
Embark’s November 1 audio documentary reopened an argument some fans thought was fading. During the video, audio director Bence Pajor explains that recording every possible version of a contextual callout would be impractical. The game changes frequently, and new ARC Raiders items or features can require more lines. He remarks that actors “would just go crazy” if asked to read every variation.
That wording drew extensive coverage and renewed player debate. Critics pointed out that recording dialogue, scheduling additional sessions and revising scripts are ordinary parts of game development. To them, the explanation sounded uncomfortably like a justification for giving actors less work.
Others are willing to draw a line between character performances and functional callouts. A ping that identifies an item or enemy is different from a scene meant to make players care about Celeste or Shani. Some players accept licensed synthetic speech for rapidly changing notifications while expecting humans to perform the story.
The frustration comes partly from timing. Players had recently been praising improved voice acting. Hearing Embark defend TTS again can feel like a reversal, even though the studio has not announced that it is undoing the recordings.
Celeste’s Actor Has Mixed Feelings
Diana Gardner, who plays Celeste, offers a perspective that is harder to fit into either side of the argument. In the documentary, she says the arrangement gave her more freedom to take other acting jobs. If Embark only needed a few short callouts, she would not necessarily have to cancel another production to return to the studio.
She also says her original contract seemed fair, though she did not know what to expect when she finally heard the resulting voice. That experience made her question whether she could fully claim the performance as her own.
Gardner has since recorded more dialogue and performed motion capture for the Frozen Trail trailer. Some players have praised Celeste’s newer scenes for sounding warmer and more expressive. Those reactions offer a useful comparison, though listeners cannot reliably identify every AI-generated line simply by how flat or natural it sounds.
Embark has said it compensates performers for recording work and approved voice licensing. Gardner’s comments suggest an actor can consider a contract fair while still feeling uneasy about how a generated version of their performance is experienced by players.
Embark’s Reasons for Defending TTS
Pajor’s argument is strongest when applied to the ping system. Players can mark objects, enemies and locations without using a microphone. Each new object or feature can create additional combinations of words, and those lines may need revision while developers continue changing the game.
Embark recorded actors and used those performances to train a voice model. The studio could then test dialogue or generate contextual lines without arranging a new recording session every time something changed. The same flexibility helped when quest writing was still being revised close to launch.
There is a meaningful difference between using generated speech during development and leaving it in the released game. A temporary line can help a writer test a quest. A finished character scene needs a convincing performance, particularly when that dialogue is supposed to create an emotional connection.
Embark’s leadership recognizes that quality gap. In March 2026, Söderlund said a professional actor delivers a better performance than AI. In the November documentary, developers also acknowledged fair criticism and said they wanted to give performers more space.
Those statements support the view that Embark is making changes in response to player feedback. They do not amount to a promise to abandon TTS.
Human Re-Recordings and Future Quests
The shift toward actors began months before the latest controversy. With more of the game’s dialogue and features now settled, Embark has been able to expand its recording work and improve character presentation, including lip sync and additional dialogue.
For players concerned about future updates, there is still a gap in the studio’s explanation. Embark has not clearly promised that every new quest will launch with human-recorded dialogue, nor has it published a breakdown of how much generated speech remains across the current game.
The game also includes a microphone voice-masking filter for players, but that is a separate feature from generated NPC performances and contextual callouts.
Unresolved AI Voice Concerns
The studio has explained why it adopted AI voice technology and confirmed substantial re-recording work. Söderlund’s latest comments offer further reassurance that the lines criticized for poor quality have been replaced.
Players still need a more precise answer about which voices they hear in the live game. A clear policy for future quests and updated ping dialogue would also help distinguish temporary production tools from voices intended for final release.
For now, players can recognize the improvements in Celeste and other recorded performances without pretending the wider issue has been settled. Embark is putting more human acting into ARC Raiders while continuing to defend TTS for flexible systems. Until it explains exactly where that technology remains, the renewed frustration is unlikely to disappear.