Facebook

Meta Launches AI Tool “AudioCraft” To Create Sounds And Music

After a short beta phase, Meta has now released “AudioCraft”, a new AI tool that allows users to create music, sound effects, and other audio content based on text input. The group has bought or licensed extra content for training the AI.

MusicGen for music and AudioGen for sound effects

The system consists of three different AI models trained on thousands of hours of audio data. The components include MusicGen for generating music, AudioGen for generating sound effects, and the general EnCodec responsible for training the models.

MusicGen can create instrumental music of various genres based on text presets that describe mood, instruments, tempo, and other characteristics. AudioGen creates sound effects such as animal sounds, weather, mechanical sounds, and more from text descriptions.

According to Meta, AudioCraft is able to produce high-quality audio and is also easy to use.

AudioCraft is “Open Source”

The models could be useful for game developers in creating sound effects and for marketing teams in creating commercial soundtracks or effects. Meta has confirmed that it also wants to offer these AI tools as open source.

Meta is in the process of introducing or releasing one AI at a time – just recently there was a report that AI “personalities” should make Facebook cool. Meta wants to create chatbots that offer new content and direct conversations, for example as well-known personalities like Abraham Lincoln.