Sound designers launch alliance to fight AI scraping and unlicensed use
The footsteps, door slams and explosions you hear in films and games now have an industry body of their own.
Sound designers, recordists, mixers and sound-effects library companies have formed the Professional Sound Alliance to fight unlicensed use of professional sound in media production and artificial intelligence, Music Business Worldwide reported.
The group dates its launch to Thursday, September 24. Its chair is Douglas Price, founder and chief executive of Pro Sound Effects. Its coordinator is Jeremy Siegel, director of strategic partnerships at the same company.
The alliance defines professional sound in plain terms. It is everything heard in a film, game or television show that is neither music nor voice. That includes ambiences, footsteps, door slams and explosions.
The fight is on two fronts. The group says it was formed in response to what it calls Data Piracy and Sync Piracy.
Data Piracy, in its definition, means scraping sound libraries into commercial AI models without permission or payment. Sync Piracy, a term that refers to the licensing of audio for use with picture, means use of sounds and libraries in film, television and games without a licence.
The alliance says it will push for fair licensing of sound design and sound effects across film, television, games and AI training. Licensing here means paid permission to use a recording, with terms set by the rights holder.
Its core group mixes library executives and working mixers. It includes Pierre Langer, chief executive of Boom Library, and Craig Nimens, vice president of Sound Ideas.
It also includes Karol Urban, an Emmy-nominated re-recording mixer, the person who balances dialogue, effects and music for the final soundtrack, and a former president of the Cinema Audio Society. The other named member is Mark Mangini, a two-time Academy Award winner whose credits include Mad Max: Fury Road, Dune and Blade Runner 2049.
The wider argument, set out on the group's website, is that every film, game, broadcast, video and AI model rests on three types of audio: music, voice and everything else. Music, it says, is represented by labels, the RIAA and performing-rights bodies. Voice is represented by SAG-AFTRA and performers' unions. Sound, it argues, has lacked equivalent cover. Professional Sound Alliance
There is already a licensed route for AI developers, at least in one catalogue. Pro Sound Effects sells an AI training dataset of 1.27 million sounds, described as rights-cleared and human-tagged, totalling more than 5,000 hours of audio.
What makes this stand out for viewers and players is the invisibility of the work. A score gets a credit and a voice gets a union card. A rainstorm, a crowd or a lightsaber hum often passes unnoticed. This group wants that third category counted and paid.
