Blizzard Entertainment has announced that audio transcription for reported voice chat has been rolled out to some Overwatch 2 players.
The November 17th Patch Notes for Overwatch 2 explain that audio transcription will begin as part of the “Defense Matrix” initiative; previously revealed to include recording of player voice-chats and requiring their phone number.
While the latter has been implemented in Call of Duty: Modern Warfare II, “newly created Call of Duty: Modern Warfare accounts,” and other free-to-play games, it was still part of the reasons players were outraged with the game on launch.
“The audio transcription feature has been added with a limited rollout for PC players in select countries,” Blizzard Entertainment explains.
“As a piece of our Defense Matrix initiative, audio transcriptions allow us to analyze a transcript from a temporary voice chat recording of a reported player. Players will now see a notification when entering voice chat for the first time during a play session indicating that voice chat may be recorded,” Blizzard Entertainment detail.
“Make sure you report as close to when disruptive behavior occurred to maximize this feature’s effectiveness,” Blizzard Entertainment encouraged. “Once reported, a temporary audio recording will be used to make a text file transcript through speech-to-text programs.”
“No one listens to the temporary audio recording, which is quickly deleted after being transcribed,” Blizzard Entertainment reveals, meaning any human element in the moderating process may need to purely rely on AI transcripted audio being accurate.
“Transcriptions are only made for reports of disruptive behavior in public voice channels, which includes team and match chat—group ‘party’ chat is excluded.” The rest of the Patch Notes continue, discussing changes to Aim Assist in Crossplay lobbies, tweaks to Heroes, and general bug fixes.
AI based moderation has been a talking point for a large chunk of 2022. Most recently parent company Activision Blizzard announced they were seeking out AI moderation to fend off hate-speech and discrimination in their games.
This was seemingly motivated by US senators asking major gaming companies about their moderation, after a damning (but potentially flawed) report by the Anti-Defamation League.
Ubisoft and Riot Games recently announced a joint effort for training AI to preemptively halt “disruptive behavior.” Even so, AI based moderation is hardly flawless. Apex Legends caught flack in prior years for banning Japanese players for saying “nigero”- Japanese for “run.” Players using other languages have also been banned for words sounding too close to slurs.