Efficient Training Data Generation for Phase-Based DOA Estimation

TL;DR

This paper introduces a low complexity online data generation method for training deep learning models in phase-based DOA estimation, reducing data storage and generation time while maintaining performance.

Contribution

It presents a novel deterministic and statistical modeling approach for efficient phase-based data generation in DOA estimation training.

Findings

Model trained with proposed data performs comparably to source-image based data models.

Reduces storage and computation requirements for training data.

Effective with measured room impulse responses.

Abstract

Deep learning (DL) based direction of arrival (DOA) estimation is an active research topic and currently represents the state-of-the-art. Usually, DL-based DOA estimators are trained with recorded data or computationally expensive generated data. Both data types require significant storage and excessive time to, respectively, record or generate. We propose a low complexity online data generation method to train DL models with a phase-based feature input. The data generation method models the phases of the microphone signals in the frequency domain by employing a deterministic model for the direct path and a statistical model for the late reverberation of the room transfer function. By an evaluation using data from measured room impulse responses, we demonstrate that a model trained with the proposed training data generation method performs comparably to models trained with data generated based on the source-image method.