Roadmap towards Superhuman Speech Understanding using Large Language Models

TL;DR

This paper presents a comprehensive five-level roadmap for developing superhuman speech understanding using large language models, introduces a benchmark for evaluation, and discusses current limitations and future directions.

Contribution

It proposes a structured roadmap for speech LLM development, introduces the SAGI Benchmark for standardized evaluation, and analyzes current challenges in handling non-semantic and acoustic information.

Findings

Gaps in handling paralinguistic cues identified

Challenges in integrating abstract acoustic knowledge highlighted

Benchmark reveals limitations in current speech LLM capabilities

Abstract

The success of large language models (LLMs) has prompted efforts to integrate speech and audio data, aiming to create general foundation models capable of processing both textual and non-textual inputs. Recent advances, such as GPT-4o, highlight the potential for end-to-end speech LLMs, which preserves non-semantic information and world knowledge for deeper speech understanding. To guide the development of speech LLMs, we propose a five-level roadmap, ranging from basic automatic speech recognition (ASR) to advanced superhuman models capable of integrating non-semantic information with abstract acoustic knowledge for complex tasks. Moreover, we design a benchmark, SAGI Bechmark, that standardizes critical aspects across various tasks in these five levels, uncovering challenges in using abstract acoustic knowledge and completeness of capability. Our findings reveal gaps in handling paralinguistic cues and abstract acoustic knowledge, and we offer future directions. This paper outlines a roadmap for advancing speech LLMs, introduces a benchmark for evaluation, and provides key insights into their current limitations and potential.