Steered Response Power for Sound Source Localization: a tutorial review

Area of Science:

• Signal Processing
• Acoustics
• Computational Audition

Background:

• The Steered Response Power (SRP) method has been a cornerstone of Sound Source Localization (SSL) for over 30 years.
• Its effectiveness in moderately reverberant and noisy conditions has led to numerous extensions.
• These extensions aim to enhance computational efficiency, enable multi-source localization, and improve performance in challenging acoustic environments.

Purpose of the Study:

• To provide a comprehensive review of the Steered Response Power (SRP) method and its variants, with a specific focus on SRP-PHAT.
• To introduce eXtensible-SRP (X-SRP), a novel, generalized, and modularized framework for SRP algorithms.
• To facilitate the implementation of existing and future SRP extensions within a unified structure.

Main Methods:

• Systematic literature review of over 200 papers on SRP and its variants.
• Development of the eXtensible-SRP (X-SRP) algorithm, a generalized and modular SRP framework.
• Implementation of X-SRP in Python, incorporating selected extensions from the reviewed literature.

Main Results:

• A comprehensive overview of the evolution and diversification of the SRP method for Sound Source Localization.
• The introduction of X-SRP, a flexible and extensible platform for SRP-based SSL.
• A functional Python implementation of X-SRP with selected literature extensions, enabling practical application and further research.

Conclusions:

• The SRP method remains a vital tool for Sound Source Localization, with ongoing research addressing its limitations.
• X-SRP offers a standardized and adaptable approach to implementing and evaluating SRP variants.
• The provided Python implementation serves as a valuable resource for researchers in acoustic signal processing and sound source localization.