次世代ネットワークにおける超臨場感音響相互通信の実現 -Ultra-realistic acoustic interactive communication on next-generation Internet-

Special Interest Workshop

The Second Workshop of A3 Foresight Program

The 2nd workshop was held with the following program.

Date: February 5-7, 2015
Workshop Venue: Hoam Faculty House, Seoul National University, Korea

5 February (Thursday)
16:00-18:00: PI meeting
19:00-20:00: Welcome reception

6 February (Friday)
10:00-11:50: Session 1 Machine learning approaches to speech and audio processing (1)

10:00-10:25 (25 min): Speech understanding in noise across multiple languages (Junfeng Li)
10:25-10:50 (25 min): Estimation of acoustic features by adaptive-network-based fuzzy inference system to synthesize emotional speech in valence-activation space (Hamada Yasuhiro, Reda Elbarougy, and Masato Akagi)
10:50:11:00 (10 min): Coffee Break
11:00-11:25 (25 min): Query-by-example spoken term detection using DTW (Ge Zhang)
11:25-11:50 (25 min): Distributed training of acoustic model with two-stage asynchronous stochastic gradient descent method (Zhichao Wang)

13:30-13:55 (25 min): NMF-based target source separation using deep learning network (Tae Gyun Kang)
13:55-14:20 (25 min): Acoustic event classification in noisy environment (Inkyu Choi)
14:20-14:45 (25 min): Support vector machine-based speaker identification employing ambient noise awareness (Inyoung Hwang)
14:45-15:00 (15 min): Coffee break

15:00-16:00 (60 min): Invited Lecture ? Audio feature learning for music information research (Juhan Nam)
16:00-16:15 (15 min): Coffee Break
16:15-16:40 (25 min): Language dependency of singing enthusiasm (Hao Zeyu)
16:40-17:05 (25 min): Melody transcription by analyzing homogeneity in the harmonic structure (Hoon Heo)
17:05-17:30 (25 min): Monaural informed source separation with limited time-frequency annotation (Il-Young Jeong)