LightNews — News, ranked by scholars

Sevinç Müjde (Çakıcı)

@sevincm.bsky.social

Çay güzelmiş 😊☕️

#YeşilBursa

January 17, 2026 at 5:36 AM

Deniz

@deniz2620.bsky.social

#BursaYanıyor gerçekten Bursa’nın yeşilinin tonunu bilseniz keşke! Duydunuz belki ama inanamazsınız görseniz! Ve Bursa siyah şimdi …🥺 #yeşilbursa kalbim acıyor 🤕

July 28, 2025 at 9:43 PM

arXiv Sound

@arxiv-sound.bsky.social

PASE, a generative speech enhancer leveraging WavLM, uses representation distillation to mitigate linguistic hallucinations by cleaning final-layer features and reduces acoustic hallucinations with a dual-stream vocoder.

PASE: Leveraging the Phonological Prior of WavLM for Low-Hallucination Generative Speech Enhancement

Xiaobin Rong, Qinwen Hu, Mansur Yesilbursa, Kamil Wojcicki, Jing Lu

arxiv.org

November 18, 2025 at 11:08 AM

arXiv cs.SD Sound

@cssd-bot.bsky.social

Wojcicki, Isik, Lechler, Yesilbursa, Bali\'c, Mack, {\L}aganowski, Zhang, Adi, Kim, Watanabe: Low-Resource Audio Codec (LRAC): 2025 Challenge Description https://arxiv.org/abs/2510.23312 https://arxiv.org/pdf/2510.23312 https://arxiv.org/html/2510.23312

October 28, 2025 at 6:34 AM

arXiv eess.AS Audio and Speech Processing

@eessas-bot.bsky.social

Xiaobin Rong, Jun Gao, Zheng Wang, Mansur Yesilbursa, Kamil Wojcicki, Jing Lu: StuPASE: Towards Low-Hallucination Studio-Quality Generative Speech Enhancement https://arxiv.org/abs/2603.09234 https://arxiv.org/pdf/2603.09234 https://arxiv.org/html/2603.09234

March 11, 2026 at 6:35 AM

arXiv eess.AS Audio and Speech Processing

@eessas-bot.bsky.social

Xiaobin Rong, Qinwen Hu, Mansur Yesilbursa, Kamil Wojcicki, Jing Lu: PASE: Leveraging the Phonological Prior of WavLM for Low-Hallucination Generative Speech Enhancement https://arxiv.org/abs/2511.13300 https://arxiv.org/pdf/2511.13300 https://arxiv.org/html/2511.13300

November 18, 2025 at 6:35 AM

arXiv eess.AS Audio and Speech Processing

@eessas-bot.bsky.social

Mack, Topaloglu, Lechler, Bali\'c, Craciun, Yesilbursa, Wojcicki: Assessing speech quality metrics for evaluation of neural audio codecs under clean speech conditions https://arxiv.org/abs/2509.24457 https://arxiv.org/pdf/2509.24457 https://arxiv.org/html/2509.24457

September 30, 2025 at 6:35 AM

arXiv Sound

@arxiv-sound.bsky.social

The 2025 Low-Resource Audio Codec Challenge targets the development of neural codecs for resource-constrained applications, focusing on robustness and integration with speech enhancement.

Low-Resource Audio Codec (LRAC): 2025 Challenge Description

Kamil Wojcicki, Yusuf Ziya Isik, Laura Lechler, Mansur Yesilbursa, Ivana Balić, Wolfgang Mack, Rafał Łaganowski, Guoqing Zhang, Yossi Adi, Minje Kim, Shinji Watanabe

arxiv.org

October 28, 2025 at 10:54 AM

arXiv Sound

@arxiv-sound.bsky.social

45 objective speech-quality metrics were evaluated by correlating their scores with subjective listening scores for clean speech across 17 codec conditions; neural-based metrics achieved the highest Pearson correlations; non-intrusive metrics tend to saturate at high subjective quality levels.

Assessing speech quality metrics for evaluation of neural audio codecs under clean speech conditions

Wolfgang Mack, Nezih Topaloglu, Laura Lechler, Ivana Balić, Alexandra Craciun, Mansur Yesilbursa, Kamil Wojcicki

arxiv.org

September 30, 2025 at 10:13 AM