Comparison of Parameter Estimation Methods for Single-Microphone Multi -Frame Wiener Filtering

Fischer, Dorte; Brumann, Klaus; Doclo, Simon

doi:10.23919/eusipco.2019.8902974

Cited by 3 publications

(2 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Requirements: To ensure the real-time operation of these models on typical hardware available today, the processing mode of the models must satisfy the following constraints on the overall introduced latency being equal to or lower than 20 ms. We define and give examples for algorithmic and buffering latency on the challenge website 4 . The realtime factor, measured as execution time on an Intel Core i5 Quadcore clocked at 2.4 GHz using single threading, must be less than 0.5.…”

Section: Challenge Tracksmentioning

confidence: 99%

“…Common DSPbased approaches for noise suppression are mostly based on spectral suppression rules such as Wiener filtering or log-short-time spectral amplitude estimators [1], [2] and often model only stationary background noise [3]. More advanced techniques include multi-frame filtering [4]. Reverberation has to be modeled in a separate estimation module and combined suppression is not straightforward [5].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

ICASSP 2023 Deep Noise Suppression Challenge

Dubey,

Aazami,

Gopal

et al. 2024

IEEE Open J. Signal Process.

View full text Add to dashboard Cite

The ICASSP 2023 Deep Noise Suppression (DNS) Challenge marks the fifth edition of the DNS challenge series. DNS challenges were organized from 2019 to 2023 to foster research in the field of DNS. Previous DNS challenges were held at INTERSPEECH 2020, ICASSP 2021, INTERSPEECH 2021, and ICASSP 2022 This challenge aims to advance models capable of jointly addressing denoising, dereverberation, and interfering talker suppression, with separate tracks focusing on headset and speakerphone scenarios. The challenge facilitates personalized deep noise suppression by providing accompanying enrollment clips for each test clip, each containing the primary talker only, which can be used to compute a speaker identity feature and disentangle primary and interfering speech. While the majority of models submitted to the challenge were personalized, the same teams emerged as the winners in both tracks. The best models demonstrated improvements of 0.145 and 0.141 in the challenge's score, respectively, when compared to the noisy blind test set. We present additional analysis and draw comparisons to previous challenges.

show abstract

Section: Challenge Tracksmentioning

confidence: 99%