#Audio, Speech & NLP

At EMNLP 2023, the Top Conference in the Field of Natural Language Processing, Sony’s R&D activities were presented.

Dialogue Generation Conditional on Predefined Stories: Preliminary Results

Self-supervised audio encoder with contrastive pretraining for Respiratory Anomaly Detection

Distortion Audio Effects: Learning How to Recover the Clean Signal

Automatic Music Mixing with Deep Learning and Out-of-Domain Data

Secondary channel estimation in spatial active noise control systems using a single moving higher order microphone

360 Virtual Mixing Environment

Fast Convergent Method for Active Noise Control Over Spatial Region with Causal Constraint

Hierarchical disentangled representation learning for singing voice conversion

Sound directivity by PT-symmetric acoustic dipoles

PT-symmetric Helmholtz resonator dipoles for sound directivity

Sound Quality Improvement of MPEG-H 3D Audio Encoder

Similarity-and-Independence-Aware Beamformer: Method for Target Source Extraction using Magnitude Spectrogram as Reference

Analytic error control methods for efficient rotation in dynamic binaural rendering of Ambisonics

Directional Dependency of Subjective Sound Pressure Perception on Three-Dimensional Sound

Metric Learning with Background Noise Class for Few-shot Detection of Rare Sound Events

Improving Voice Separation by Incorporating End-To-End Speech Recognition

Array-Geometry-Aware Spatial Active Noise Control Based on Direction-of-Arrival Weighting