AudioLDM 2: Learning Holistic Audio Generation with Self-supervised Pretraining

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Liu, Haohe, Yuan, Yi, Liu, Xubo, Mei, Xinhao, Kong, Qiuqiang, Tian, Qiao, Wang, Yuping, Wang, Wenwu, Wang, Yuxuan, Plumbley, Mark D.
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913347337715712
author Liu, Haohe
Yuan, Yi
Liu, Xubo
Mei, Xinhao
Kong, Qiuqiang
Tian, Qiao
Wang, Yuping
Wang, Wenwu
Wang, Yuxuan
Plumbley, Mark D.
author_facet Liu, Haohe
Yuan, Yi
Liu, Xubo
Mei, Xinhao
Kong, Qiuqiang
Tian, Qiao
Wang, Yuping
Wang, Wenwu
Wang, Yuxuan
Plumbley, Mark D.
contents Although audio generation shares commonalities across different types of audio, such as speech, music, and sound effects, designing models for each type requires careful consideration of specific objectives and biases that can significantly differ from those of other types. To bring us closer to a unified perspective of audio generation, this paper proposes a framework that utilizes the same learning method for speech, music, and sound effect generation. Our framework introduces a general representation of audio, called "language of audio" (LOA). Any audio can be translated into LOA based on AudioMAE, a self-supervised pre-trained representation learning model. In the generation process, we translate any modalities into LOA by using a GPT-2 model, and we perform self-supervised audio generation learning with a latent diffusion model conditioned on LOA. The proposed framework naturally brings advantages such as in-context learning abilities and reusable self-supervised pretrained AudioMAE and latent diffusion models. Experiments on the major benchmarks of text-to-audio, text-to-music, and text-to-speech demonstrate state-of-the-art or competitive performance against previous approaches. Our code, pretrained model, and demo are available at https://audioldm.github.io/audioldm2.
format Preprint
id arxiv_https___arxiv_org_abs_2308_05734
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle AudioLDM 2: Learning Holistic Audio Generation with Self-supervised Pretraining
Liu, Haohe
Yuan, Yi
Liu, Xubo
Mei, Xinhao
Kong, Qiuqiang
Tian, Qiao
Wang, Yuping
Wang, Wenwu
Wang, Yuxuan
Plumbley, Mark D.
Sound
Artificial Intelligence
Multimedia
Audio and Speech Processing
Signal Processing
Although audio generation shares commonalities across different types of audio, such as speech, music, and sound effects, designing models for each type requires careful consideration of specific objectives and biases that can significantly differ from those of other types. To bring us closer to a unified perspective of audio generation, this paper proposes a framework that utilizes the same learning method for speech, music, and sound effect generation. Our framework introduces a general representation of audio, called "language of audio" (LOA). Any audio can be translated into LOA based on AudioMAE, a self-supervised pre-trained representation learning model. In the generation process, we translate any modalities into LOA by using a GPT-2 model, and we perform self-supervised audio generation learning with a latent diffusion model conditioned on LOA. The proposed framework naturally brings advantages such as in-context learning abilities and reusable self-supervised pretrained AudioMAE and latent diffusion models. Experiments on the major benchmarks of text-to-audio, text-to-music, and text-to-speech demonstrate state-of-the-art or competitive performance against previous approaches. Our code, pretrained model, and demo are available at https://audioldm.github.io/audioldm2.
title AudioLDM 2: Learning Holistic Audio Generation with Self-supervised Pretraining
topic Sound
Artificial Intelligence
Multimedia
Audio and Speech Processing
Signal Processing
url https://arxiv.org/abs/2308.05734