Meta Platforms has introduced an open-source AI tool called “AudioCraft” that enables users to generate music and audio based on text prompts. The tool includes three models, including MusicGen trained on proprietary and licensed music, but concerns about copyright violations arise as such AI tools rely on web-sourced data for pattern recognition.
2 August 2023 – Meta Platforms has unveiled an open-source AI tool named “AudioCraft,” designed to assist users in crafting music and audio content based on textual prompts. The tool features three models—AudioGen, EnCodec, and MusicGen—enabling the creation of music, sound, compression, and generation, with MusicGen being trained using Meta’s proprietary and licensed music resources.
While the innovation holds promise for music and audio enthusiasts, concerns about potential copyright infringements arise due to machine learning algorithms drawing patterns from web-sourced data. This move comes in the wake of Alphabet’s introduction of its experimental AI audio generation tool, MusicLM, earlier this year. – source: Reuters
