Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
102 tokens/sec
GPT-4o
59 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
50 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Can Large Language Models Understand Spatial Audio? (2406.07914v2)

Published 12 Jun 2024 in cs.SD and eess.AS

Abstract: This paper explores enabling LLMs to understand spatial information from multichannel audio, a skill currently lacking in auditory LLMs. By leveraging LLMs' advanced cognitive and inferential abilities, the aim is to enhance understanding of 3D environments via audio. We study 3 spatial audio tasks: sound source localization (SSL), far-field speech recognition (FSR), and localisation-informed speech extraction (LSE), achieving notable progress in each task. For SSL, our approach achieves an MAE of $2.70{\circ}$ on the Spatial LibriSpeech dataset, substantially surpassing the prior benchmark of about $6.60{\circ}$. Moreover, our model can employ spatial cues to improve FSR accuracy and execute LSE by selectively attending to sounds originating from a specified direction via text prompts, even amidst overlapping speech. These findings highlight the potential of adapting LLMs to grasp physical audio concepts, paving the way for LLM-based agents in 3D environments.

User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (11)
  1. Changli Tang (15 papers)
  2. Wenyi Yu (14 papers)
  3. Guangzhi Sun (51 papers)
  4. Xianzhao Chen (10 papers)
  5. Tian Tan (21 papers)
  6. Wei Li (1122 papers)
  7. Jun Zhang (1008 papers)
  8. Lu Lu (189 papers)
  9. Zejun Ma (78 papers)
  10. Yuxuan Wang (239 papers)
  11. Chao Zhang (907 papers)
Citations (2)
X Twitter Logo Streamline Icon: https://streamlinehq.com
Reddit Logo Streamline Icon: https://streamlinehq.com