In Amazon’s game engine, voice actors can now be replaced with robots
Want to add voice acting to your next epic video game but don’t want to deal with those pesky real-life actors to populate your virtual towns and castles? Amazon has your money-saving back.
The company’s Lumberyard game engine now supports a full text-to-speech pipeline in its 1.11 version, which is now live for any of its developers. A demonstration video shows how built-in tools allow game developers to attach text to any interaction in a game, which can be spoken in one of 50 “voices” in 24 different languages. What’s more, the engine’s toolset will also automatically render a lip-synced animation for any voiced 3D characters in your game project.
Amazon’s brief demo video of the feature only includes a select few voice samples and a very brief demonstration of the lip sync feature, which looks serviceable but limited. (For a comparison point, it looks about as so-so as, say, the system in 2011’s Elder Scrolls V: Skyrim.) In the case of the latter, Amazon showed a character with separately animated facial and eye systems, which may obscure Lumberyard’s automatic lip-sync capabilities.
The industry of video game engines has exploded in recent years, with engines like Unity and Unreal catching a ton of market share. Other game engines haven’t fared as well in the latest wave, however, and the highest-profile example is probably Amazon’s Lumberyard—a repurposed CryEngine for all intents and purposes. It launched in 2016 with a competitive royalty-free structure, which potentially reduces its up-front cost to game makers. However, in addition to adopting some of CryEngine’s issues as a game-development platform, particularly uneven documentation, Amazon’s option requires that its game makers attach any Web services to the company’s paid AWS platform. For some online indies, that requirement can ultimately prove cost-prohibitive.
Coincidentally, this feature announcement arrived the same day that the United States’ largest voice-actor union, SAG-AFTRA, announced the official end of its strike on video game projects. That strike, which began in October of last year, had been the longest ever by a group of actors, and it impacted more than a few video games’ casting decisions. However, we’ve yet to see a major game opt to use computer-generated speech for its character cast. While Amazon’s “Polly” sample sounds serviceable enough, it doesn’t reach the same natural speech patterns of its Alexa voice assistant, nor does it compare to Google’s promising WaveNet platform. (WaveNet has yet to be released or licensed for video game projects.)
Author Sam Machkovech