EUVIMEDEUVIMED entdecken
Crossref

Self-supervised speech quality assessment (S3QA): Leveraging speech foundation models for a scalable speech quality metric

Mattson Ogg iD, Caitlyn A Bishop, Han Gyol Yi, Sarah R Robinson

The Journal of the Acoustical Society of America · 01.04.2026 · 159 (4) · 3662-3673

DOI: 10.1121/10.0043367 ↗

Kurz gefasst

Fachartikel in der Zeitschrift The Journal of the Acoustical Society of America, online erschienen am 27.04.2026. Verfasst von Mattson Ogg und 3 weiteren Autor:innen. Beteiligte Einrichtung: Johns Hopkins University Applied Physics Laboratory, Laurel, Maryland 20723, USA. Hauptthemen laut MEDLINE: Speech Acoustics, Speech Production Measurement, Voice Quality, Speech Intelligibility. Die Arbeit nennt 68 Literaturangaben. Gefördert von: NIDA NIH HHS.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Methods for automatically assessing speech quality in real world environments are critical for robust human language technologies and assistive devices. Behavioral ratings (e.g., mean opinion scores) are considered the gold standard, but are susceptible to inter-rater variability, cannot be generalized, and are labor-intensive, thus limiting the acoustic challenges they can quantify. We present a scalable method for assessing speech quality: the self-supervised speech quality assessment model. First, we manipulated high-quality utterances, using acoustic challenges that emulate real-world degradation: filtering, reverberation, background noise, and digital compression. Second, we leveraged a pre-trained speech foundation model, WavLM, to derive cosine distances between the clean and degraded versions of each utterance in the embedding space. A transformer-based model was trained to predict these cosine distances, given only the degraded utterances. The trained model was evaluated on unseen corpora of synthetic mixtures, NISQA, and VOiCES. The self-supervised speech quality assessment model accurately predicts degradation cosine distances across a wide range of acoustic challenges and is aligned with behavioral ratings (mean opinion scores), automatic speech recognition performance and other important features (microphone distances). The model is available online.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Speech AcousticsSpeech Production MeasurementVoice QualitySpeech IntelligibilitySignal Processing, Computer-AssistedHumansAcousticsNoise

Autor:innen und Institutionen

  • Johns Hopkins University Applied Physics Laboratory, Laurel, Maryland 20723, USA.

ORCID-Profile: Mattson Ogg

Förderung

  • NIDA NIH HHS

Zitieren

APA

Ogg, M., Bishop, C. A., Yi, H. G., & Robinson, S. R. (2026). Self-supervised speech quality assessment (S3QA): Leveraging speech foundation models for a scalable speech quality metric. The Journal of the Acoustical Society of America, 159(4), 3662-3673. https://doi.org/10.1121/10.0043367

Vancouver

Ogg M, Bishop CA, Yi HG, Robinson SR. Self-supervised speech quality assessment (S3QA): Leveraging speech foundation models for a scalable speech quality metric. The Journal of the Acoustical Society of America. 2026;159(4):3662-3673. doi:10.1121/10.0043367

BibTeX
@article{Ogg2026,
  author = {Ogg, Mattson and Bishop, Caitlyn A and Yi, Han Gyol and Robinson, Sarah R},
  title = {Self-supervised speech quality assessment (S3QA): Leveraging speech foundation models for a scalable speech quality metric},
  journal = {The Journal of the Acoustical Society of America},
  year = {2026},
  volume = {159},
  number = {4},
  pages = {3662--3673},
  doi = {10.1121/10.0043367}
}
Kostenlos starten →