.png "blank Coverage")
Meta, the parent company of Facebook, has unveiled a new open-source AI tool called AudioCraft, designed to enable both professional musicians and everyday users to create audio and music from simple text prompts. This tool comprises three models: MusicGen, AudioGen, and EnCodec. MusicGen can generate music from text inputs, while AudioGen can generate audio based on text inputs, drawing from public sound effects. The EnCodec decoder has been improved to generate higher-quality music with fewer unwanted artifacts.
Meta is also making its pre-trained AudioGen models available to users, allowing them to generate environmental sounds and sound effects like dogs barking, cars honking, or footsteps on a wooden floor. The company is sharing all model weights and code for the AudioCraft tool, making it a valuable resource for various applications, including music composition, sound effects generation, compression algorithms, and audio generation.
According to Meta, generative AI has made significant strides in images, video, and text, but audio has lagged behind. AudioCraft aims to address this gap by providing an accessible and user-friendly platform for generating high-quality audio. Meta's decision to open-source these models will enable researchers and practitioners to train their own models using their datasets and potentially advance the field of audio generation.
Creating realistic and high-fidelity audio is a complex task that involves modeling intricate signals and patterns at different scales. Music, in particular, presents a unique challenge due to its composition of local and long-range patterns. AudioCraft simplifies the design of generative models for audio, making it easier for users to experiment with existing models and generate audio over long durations. This tool has the potential to democratize audio creation and innovation.
