NHL 27 Replaces Studio Time With AI Voice Cloning for In-Game Commentary

The voice calling the action in the September 11 release of NHL 27 is not entirely human. Veteran ESPN broadcaster John Buccigross confirmed that EA Sports utilized generative AI to synthesize his voice for the game's play-by-play commentary, fundamentally altering how the 2026 sports title handles dynamic audio.[1]

Instead of spending hundreds of hours in a recording booth reading thousands of player names and situational permutations, Buccigross provided a baseline voice model. The game's engine then generates specific lines on the fly to match the on-ice action. Speaking on the Chirping Zebras podcast, Buccigross noted the technology allowed him to do less physical work while retaining 100 percent of his standard fee.[1]

The implementation represents a major shift in the economics of sports game development. Titles across the publisher's portfolio require massive audio databases to account for every possible roster move and statistical milestone. By cloning a broadcaster's voice, a studio can theoretically synthesize commentary for user-created players and hyper-specific game states that would be impossible to record manually.[2]

The trade-off for that infinite scalability is a measurable drop in natural chemistry and contextual accuracy. Buccigross admitted that during the review process, the synthesized voice yelled that the crowd was getting out of their 'chairs' instead of 'seats.' Developers had to manually correct these phrasing errors, highlighting the gap between raw voice cloning and true sports broadcasting intuition.[1]

IGN's review handed the game a 6 out of 10 score, noting that it does little to improve upon the previous 12 months of development. The outlet noted that while the core hockey mechanics remain enjoyable, the overall package feels stagnant. The reliance on algorithmic audio generation rather than organic studio sessions contributes to that sense of artificiality, as the AI struggles to replicate the spontaneous banter that defines a premium sports broadcast.[1]

The broader gaming industry is closely watching the reception of NHL 27's audio suite. Generative AI disclaimers are now appearing on the PC storefronts for 3 other major EA franchises—College Football, Madden, and FC. This comes exactly 1 year after Call of Duty faced intense community pushback for utilizing algorithms to generate cosmetic calling cards, signaling a growing tension between development efficiency and player expectations.[1]

Because NHL 27 is not distributed on Steam, it bypasses the platform's mandatory AI disclosure labels, leaving console players on the PlayStation 5 and Xbox Series X to discover the synthetic audio firsthand. The lack of upfront transparency has fueled debates on community forums regarding the value proposition of full-priced annual sports releases.[1][2]

For now, the technology serves as a functional bridge. It ensures that every obscure minor-league call-up has their name pronounced correctly, but it sacrifices the raw, unscripted energy of a live broadcast booth. The next verifiable checkpoint for the technology will be the 2027 release cycle, when developers must prove their algorithms can synthesize genuine excitement rather than just reading the correct names.[2]