Pengi: An Audio Language Model for Audio Tasks

TL;DR

Pengi is a versatile audio language model that transforms all audio tasks into text-generation problems, enabling open-ended and closed-ended tasks without additional fine-tuning, achieving state-of-the-art results across multiple benchmarks.

Contribution

Introduces Pengi, a novel Audio Language Model that unifies audio tasks as text-generation, allowing open-ended tasks without task-specific modifications.

Findings

State-of-the-art performance on 22 downstream tasks

Effective handling of open-ended audio tasks like captioning and Q&A

Unified architecture enables versatile audio understanding

Abstract

In the domain of audio processing, Transfer Learning has facilitated the rise of Self-Supervised Learning and Zero-Shot Learning techniques. These approaches have led to the development of versatile models capable of tackling a wide array of tasks, while delivering state-of-the-art performance. However, current models inherently lack the capacity to produce the requisite language for open-ended tasks, such as Audio Captioning or Audio Question & Answering. We introduce Pengi, a novel Audio Language Model that leverages Transfer Learning by framing all audio tasks as text-generation tasks. It takes as input, an audio recording, and text, and generates free-form text as output. The input audio is represented as a sequence of continuous embeddings by an audio encoder. A text encoder does the same for the corresponding text input. Both sequences are combined as a prefix to prompt a pre-trained frozen language model. The unified architecture of Pengi enables open-ended tasks and close-ended tasks without any additional fine-tuning or task-specific extensions. When evaluated on 22 downstream tasks, our approach yields state-of-the-art performance in several of them. Our results show that connecting language models with audio models is a major step towards general-purpose audio understanding