Skim: Skip** memory lstm for low-latency real-time continuous speech separation

C Li, L Yang, W Wang, Y Qian - ICASSP 2022-2022 IEEE …, 2022 - ieeexplore.ieee.org
Continuous speech separation for meeting pre-processing has recently become a focused
research topic. Compared to the data in utterance-level speech separation, the meeting …

Online binaural speech separation of moving speakers with a wavesplit network

C Han, N Mesgarani - ICASSP 2023-2023 IEEE International …, 2023 - ieeexplore.ieee.org
Binaural speech separation in real-world scenarios often involves moving speakers. Most
current speech separation methods use utterance-level permutation invariant training (u …

Segment-less continuous speech separation of meetings: Training and evaluation criteria

T von Neumann, K Kinoshita… - … on Audio, Speech …, 2022 - ieeexplore.ieee.org
Continuous Speech Separation (CSS) has been proposed to address speech overlaps
during the analysis of realistic meeting-like conversations by eliminating any overlaps before …

Dual-path modeling with memory embedding model for continuous speech separation

C Li, Z Chen, Y Qian - IEEE/ACM Transactions on Audio …, 2022 - ieeexplore.ieee.org
Continuous speech separation (CSS) aims at separating overlap-free targets from a long,
partially-overlapped recording. Though it has shown promising results, the origin CSS …

Directed speech separation for automatic speech recognition of long form conversational speech

R Paturi, S Srinivasan, K Kirchhoff… - ar** Reference Speech Estimation Method for Speaker Extraction
Y Zhang, Z Li, B Liu, H Fan, Y Yang, Q Yang - International Conference on …, 2024 - Springer
Speaker extraction is a technique that separates the target speech from multi-talker mixtures
using a priori information about the target speaker, such as pre-enrolled reference speech …

[CARTE][B] Automatic speech separation for brain-controlled hearing technologies

C Han - 2024 - search.proquest.com
Speech perception in crowded acoustic environments is particularly challenging for hearing
impaired listeners. While assistive hearing devices can suppress background noises distinct …

[PDF][PDF] OR-TSE: An Overlap-Robust Speaker Encoder for Target Speech Extraction

Y Zhang, L Yao, Q Yang - Proc. Interspeech 2024, 2024 - isca-archive.org
Abstract Mainstream Target Speech Extraction (TSE) systems extract target speech from a
mixture using pre-enrolled reference speech. The extraction performance heavily depends …

A Dual-Branch Speech Enhancement Model with Harmonic Repair

L Jia, Y Xu, D Ke - Applied Sciences, 2024 - mdpi.com
Recent speech enhancement studies have mostly focused on completely separating noise
from human voices. Due to the lack of specific structures for harmonic fitting in previous …