arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.04825cs.SDcs.LGeess.AS
收藏

Fast Timing-Conditioned Latent Audio Diffusion(基于时序条件的快速潜在音频扩散模型)

Fast Timing-Conditioned Latent Audio Diffusion

Stability AI;Belmont University(贝尔蒙特大学)

Zach Evans, CJ Carr, Josiah Taylor, Scott H. Hawley, Jordi Pons

AI总结:

本研究提出基于潜在扩散的Stable Audio模型,通过文本提示与时序嵌入实现长时长、可变长度的44.1kHz立体声音乐与音效高效生成,推理速度快且基准测试表现优异,还可生成有结构的音乐和立体声音效。

摘要:

从文本提示生成长时长44.1kHz立体声音频的计算成本通常很高。此外,以往大多数研究未解决音乐和音效时长天然存在差异的问题。本研究聚焦于使用生成模型,通过文本提示高效生成长时长、可变长度的44.1kHz立体声音乐与音效。Stable Audio基于潜在扩散技术,其潜在空间由全卷积变分自编码器定义,同时以文本提示和时序嵌入为条件,可精细控制生成音乐与音效的内容和时长。在A100 GPU上,Stable Audio能在8秒内渲染最长95秒的44.1kHz立体声信号。尽管该模型计算效率高、推理速度快,在两个公开的文本转音乐和文本转音频基准测试中仍处于顶尖水平;与现有最优模型不同,它还能生成具有结构的音乐和立体声音效。

英文摘要:

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.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑