Fast Timing-Conditioned Latent Audio Diffusion
This addresses the computational demands and lack of duration control in text-to-audio generation for music and sound effects, though it appears incremental as it builds on latent diffusion.
The paper tackles efficient generation of long-form, variable-length stereo music and sounds at 44.1kHz from text prompts, achieving up to 95 seconds of audio in 8 seconds on an A100 GPU and performing well in benchmarks.
Generating long-form 44.1kHz stereo audio from text prompts can be computationally demanding. Further, most previous works do not tackle that music and sound effects naturally vary in their duration. Our research focuses on the efficient generation of long-form, variable-length stereo music and sounds at 44.1kHz using text prompts with a generative model. Stable Audio is based on latent diffusion, with its latent defined by a fully-convolutional variational autoencoder. It is conditioned on text prompts as well as timing embeddings, allowing for fine control over both the content and length of the generated music and sounds. Stable Audio is capable of rendering stereo signals of up to 95 sec at 44.1kHz in 8 sec on an A100 GPU. Despite its compute efficiency and fast inference, it is one of the best in two public text-to-music and -audio benchmarks and, differently from state-of-the-art models, can generate music with structure and stereo sounds.