Efficient Training Data Generation for Phase-Based DOA Estimation
This work addresses a domain-specific problem for researchers and practitioners in audio signal processing by providing an incremental improvement in data generation efficiency for DOA estimation.
The paper tackled the problem of high storage and computational costs in generating training data for deep learning-based direction of arrival (DOA) estimation by proposing a low-complexity online data generation method using phase-based features, and demonstrated that models trained with this method perform comparably to those using more expensive methods.
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.