Metraipolis › archive › dc_7f7cea303fee

Towards Balanced Spectral Reconstruction: Spectrally Adaptive Loss for Streaming Speech Enhancement — arXiv:2608.30739

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.30739
License (as archived)
CC0-1.0
Archived
2026-10-17T10:42:52.655Z
SHA-256 of the archived text
b6c76a0b57eae33c0365fd79fa87eadef9f6df4f0334972804c07259d98a843b
Stable URL of this page
https://metraipolis.com/documento/dc_7f7cea303fee

How to cite

Towards Balanced Spectral Reconstruction: Spectrally Adaptive Loss for Streaming Speech Enhancement — arXiv:2608.30739 — arXiv (en), https://arxiv.org/abs/2608.30739 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7f7cea303fee (Metraipolis archive dc_7f7cea303fee, sha256 b6c76a0b57eae33c0365fd79fa87eadef9f6df4f0334972804c07259d98a843b)
BibTeX
@misc{dc7f7cea303fee,
  title = {Towards Balanced Spectral Reconstruction: Spectrally Adaptive Loss for Streaming Speech Enhancement — arXiv:2608.30739},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.30739},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7f7cea303fee; sha256 b6c76a0b57eae33c0365fd79fa87eadef9f6df4f0334972804c07259d98a843b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30739. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

This paper proposes two spectrally weighted STFT loss functions for lightweight streaming speech enhancement, addressing the magnitude over-attenuation in mid-to-high frequency regions caused by the magnitude-phase compensation effect. The proposed sigmoid-weighted loss applies a smooth frequency-dependent modulation to the phase-aware contribution, while the signal-dependent spectrally adaptive loss further conditions the modulation on the ground-truth log-magnitude spectrogram. To evaluate the proposed objectives, we additionally design HyST-Net, a lightweight and competitive backbone with hybrid MHA-GRU spectral-temporal modelling for low-latency streaming scenarios. Experimental results exhibit consistent improvements in high-frequency spectral reconstruction for both losses. The spectrally adaptive loss further enhances the mid-frequency region, resulting in a more balanced spectral reconstruction across the full frequency range.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in eess.AS).
Categorie arXiv: Audio and Speech Processing (eess.AS).
Pagina del lavoro: https://arxiv.org/abs/2608.30739 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T10:32:48.307Z dalla pagina https://arxiv.org/catchup/eess.AS/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).