Learning Audio Concepts from Counterfactual Natural Language

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Vosoughi, Ali, Bondi, Luca, Wu, Ho-Hsiang, Xu, Chenliang
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866929205972828160
author Vosoughi, Ali
Bondi, Luca
Wu, Ho-Hsiang
Xu, Chenliang
author_facet Vosoughi, Ali
Bondi, Luca
Wu, Ho-Hsiang
Xu, Chenliang
contents Conventional audio classification relied on predefined classes, lacking the ability to learn from free-form text. Recent methods unlock learning joint audio-text embeddings from raw audio-text pairs describing audio in natural language. Despite recent advancements, there is little exploration of systematic methods to train models for recognizing sound events and sources in alternative scenarios, such as distinguishing fireworks from gunshots at outdoor events in similar situations. This study introduces causal reasoning and counterfactual analysis in the audio domain. We use counterfactual instances and include them in our model across different aspects. Our model considers acoustic characteristics and sound source information from human-annotated reference texts. To validate the effectiveness of our model, we conducted pre-training utilizing multiple audio captioning datasets. We then evaluate with several common downstream tasks, demonstrating the merits of the proposed method as one of the first works leveraging counterfactual information in audio domain. Specifically, the top-1 accuracy in open-ended language-based audio retrieval task increased by more than 43%.
format Preprint
id arxiv_https___arxiv_org_abs_2401_04935
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Learning Audio Concepts from Counterfactual Natural Language
Vosoughi, Ali
Bondi, Luca
Wu, Ho-Hsiang
Xu, Chenliang
Multimedia
Computation and Language
Sound
Audio and Speech Processing
Conventional audio classification relied on predefined classes, lacking the ability to learn from free-form text. Recent methods unlock learning joint audio-text embeddings from raw audio-text pairs describing audio in natural language. Despite recent advancements, there is little exploration of systematic methods to train models for recognizing sound events and sources in alternative scenarios, such as distinguishing fireworks from gunshots at outdoor events in similar situations. This study introduces causal reasoning and counterfactual analysis in the audio domain. We use counterfactual instances and include them in our model across different aspects. Our model considers acoustic characteristics and sound source information from human-annotated reference texts. To validate the effectiveness of our model, we conducted pre-training utilizing multiple audio captioning datasets. We then evaluate with several common downstream tasks, demonstrating the merits of the proposed method as one of the first works leveraging counterfactual information in audio domain. Specifically, the top-1 accuracy in open-ended language-based audio retrieval task increased by more than 43%.
title Learning Audio Concepts from Counterfactual Natural Language
topic Multimedia
Computation and Language
Sound
Audio and Speech Processing
url https://arxiv.org/abs/2401.04935