Pseudo-Spatially Conditioned TF-Locoformer with MHCA+FiLM Fusion for Single-Channel Speech Separation
Daichi Nitsu, Koichi Shinoda
Proc. Interspeech 2026, Sep. 27- Oct. 1, 2026, Sydney, Australia, pp. 7148-7152
Improving Audio Codec-based Speech Separation by Stacking Residual Vector Quantization Layers
Nhu Minh Phuong Dinh, Roland Hartanto, Koichi Shinoda
Proc. Interspeech 2026, Sep. 27- Oct. 1, 2026, Sydney, Australia, pp. 7153-7157
Science Tokyo CHiME-9 MCoRec System Description
Roland Hartanto, Daichi Nitsu, Nhu Minh Phuong Dinh, Koichi Shinoda
Proc. 9th CHiME Speech Separation and Recognition Challenge (CHiME 2026), pp. 21-25
Identity Preserving Speaker Attribute Editing
Yuta Seki, Koichi Shinoda
Technical Report of IEICE SP, vol. 126, no. 58, pp. 31-36, 2026-5-29
Cross-layer Recurrent Processing of Residual Vector Quantization for Audio Codec-based Speech Separation
Nhu Minh Phuong, Roland Hartanto, Koichi Shinoda
Technical Report of IEICE SP, vol. 126, no. 58, pp. 125-129, 2026-5-29
Single-Channel Speech Separation with Pseudo-Spatial Embeddings
Daichi Nitsu, Koichi Shinoda
Technical Report of IEICE SP, vol. 126, no. 58, pp. 113-118, 2026-5-29
Robust Speaker Representation Learning via Self-Supervised Pre-training Using Diffusion Models
Kaito Miwa, Koichi Shinoda
Technical Report of IEICE SP, vol. 126, no. 58, pp. 61-65, 2026-5-29
Detection of Depression Using Web-Interview Data by LLM Enhanced with Multimodal Features
Isaac Morales Nolasco, Koichi Shinoda, Momoko Kitazawa, Yuriko Kaise, Shunsuke Takagi, Genichi Sugihara, and Taishiro Kishimoto
Technical Reports of IEICE PRMU, vol. 125, no. 348, pp. 49-54, January 22, 2026