MAGNET, a newly developed text-to-audio AI method by meta researchers, can swiftly generate music, sound effects, and noises from text prompts.
During training, MAGNET guesses missing parts of songs and reveals them by using context clues. This approach is 7 times speedier than comparable systems but maintains the same level of sound quality. In addition, MAGNET undergoes testing to convert text into sound effects and environmental noises.
By piecing together sections in a non-sequential manner, MAGNET points towards the prospect of enhancing AI techniques for music, audio, and potentially video creation in the future.
What do you think?
+1
+1
+1
+1
+1
