MASKED AUDIO GENERATION USING A SINGLE NONAUTOREGRESSIVE TRANSFORMER

Alon Ziv, Itai Gat, Gael Le Lan, Tal Remez, Felix Kreuk, Alexandre Défossez, Jade Copet, Gabriel Synnaeve, Yossi Adi

Research output: Contribution to conferencePaperpeer-review

Abstract

We introduce MAGNET, a masked generative sequence modeling method that operates directly over several streams of audio tokens. Unlike prior work, MAGNET is comprised of a single-stage, non-autoregressive transformer. During training, we predict spans of masked tokens obtained from a masking scheduler, while during inference we gradually construct the output sequence using several decoding steps. To further enhance the quality of the generated audio, we introduce a novel rescoring method in which, we leverage an external pretrained model to rescore and rank predictions from MAGNET, which will be then used for later decoding steps. Lastly, we explore a hybrid version of MAGNET, in which we fuse between autoregressive and non-autoregressive models to generate the first few seconds in an autoregressive manner while the rest of the sequence is being decoded in parallel. We demonstrate the efficiency of MAGNET for the task of text-to-music and text-to-audio generation and conduct an extensive empirical evaluation, considering both objective metrics and human studies. The proposed approach is comparable to the evaluated baselines, while being significantly faster (x7 faster than the autoregressive baseline). Through ablation studies and analysis, we shed light on the importance of each of the components comprising MAGNET, together with pointing to the trade-offs between autoregressive and non-autoregressive modeling, considering latency, throughput, and generation quality. Samples are available on our demo page https://pages.cs.huji.ac.il/adiyoss-lab/MAGNeT.

Original languageEnglish
StatePublished - 2024
Externally publishedYes
Event12th International Conference on Learning Representations, ICLR 2024 - Hybrid, Vienna, Austria
Duration: 7 May 202411 May 2024

Conference

Conference12th International Conference on Learning Representations, ICLR 2024
Country/TerritoryAustria
CityHybrid, Vienna
Period7/05/2411/05/24

Bibliographical note

Publisher Copyright:
© 2024 12th International Conference on Learning Representations, ICLR 2024. All rights reserved.

Fingerprint

Dive into the research topics of 'MASKED AUDIO GENERATION USING A SINGLE NONAUTOREGRESSIVE TRANSFORMER'. Together they form a unique fingerprint.

Cite this