Research

Research Projects

MangaVox comic-style dataset project banner

MangaVox

DAC architecture with SFI convolutional layers controlling neural audio token temporal resolution

Neural Audio Codecs: Variable Time Resolution and Robustness

Proposed SFI convolution layer converting latent analog filters into sampling-frequency-dependent discrete weights

Sampling-Frequency-Independent Audio Processing

Harmonic sound mixture reconstructed from two DDSP synthesizers using spectral fitting

Neural Analysis-by-Synthesis Based on DDSP Mixture Models

Contemporary A Cappella and Vocal Percussion Processing

Geodesics and decision boundaries in the Poincare ball model of hyperbolic space

Structured Audio Embeddings for Musical Timbre and Effects

Five idol-group performers singing and dancing with microphones

Japanese Idol-Group Music Corpora

Microphone-array signals processed by blockwise ILRMA and framewise RCSCME to extract target speech

Real-Time Blind Speaker Extraction

Joint neural blind source separation and speaker diarization model with spatial covariance and source activity estimation

Neural Blind Source Separation

Diagram of MRDLA multiresolution analysis and synthesis for source separation

Filterbank-Inspired Neural Analysis–Synthesis

Experimental design separating the effects of NAC and SSL pretraining languages

Speech Representation Learning and Model Analysis

Sparse acoustic observations, neural and physics-informed models, generative methods, and sound-field and HRTF interpolation

Spatial Audio Processing and Sound-Field Modeling

Three-component seismic traces separated by BSS into P-like, S-like, and residual components

Seismic Signal Decomposition Using Blind Source Separation

Diagram of harmonic-temporal factor decomposition for monaural source separation

Harmonic-Temporal Factor Decomposition

Close microphone and source placement illustrating bleeding sound between neighboring sources

Music Bleeding-Sound Reduction Based on NMF

Diagram of score following for automatic accompaniment

Score Following and Automatic Accompaniment

Illustration of spectrogram consistency in CWT magnitude reconstruction

Fast Reconstruction from Magnitude CWT Spectrograms

Illustration of timbre replacement in musical audio

Unsupervised Drum Timbre Replacement