Matches in SemOpenAlex for { <https://semopenalex.org/work/W4280557595> ?p ?o ?g. }
Showing items 1 to 85 of
85
with 100 items per page.
- W4280557595 abstract "Separation of speech mixtures in noisy and reverberant environments remains a challenging task for state-of-the-art speech separation systems. Time-domain audio speech separation networks (TasNets) are among the most commonly used network architectures for this task. TasNet models have demonstrated strong performance on typical speech separation baselines where speech is not contaminated with noise. When additive or convolutive noise is present, performance of speech separation degrades significantly. TasNets are typically constructed of an encoder network, a mask estimation network and a decoder network. The design of these networks puts the majority of the onus for enhancing the signal on the mask estimation network when used without any pre-processing of the input data or post processing of the separation network output data. Use of multihead attention (MHA) is proposed in this work as an additional layer in the encoder and decoder to help the separation network attend to encoded features that are relevant to the target speakers and conversely suppress noisy disturbances in the encoded features. As shown in this work, incorporating MHA mechanisms into the encoder network in particular leads to a consistent performance improvement across numerous quality and intelligibility metrics on a variety of acoustic conditions using the WHAMR corpus, a data-set of noisy reverberant speech mixtures. The use of MHA is also investigated in the decoder network where it is demonstrated that smaller performance improvements are consistently gained within specific model configurations. The best performing MHA models yield a mean 0.6 dB scale invariant signal-to-distortion (SISDR) improvement on noisy reverberant mixtures over a baseline 1D convolution encoder. A mean 1 dB SISDR improvement is observed on clean speech mixtures." @default.
- W4280557595 created "2022-05-22" @default.
- W4280557595 creator A5004802529 @default.
- W4280557595 creator A5027797344 @default.
- W4280557595 creator A5030528300 @default.
- W4280557595 date "2022-05-11" @default.
- W4280557595 modified "2023-10-14" @default.
- W4280557595 title "Att-TasNet: Attending to Encodings in Time-Domain Audio Speech Separation of Noisy, Reverberant Speech Mixtures" @default.
- W4280557595 cites W1603978816 @default.
- W4280557595 cites W185399533 @default.
- W4280557595 cites W2069681747 @default.
- W4280557595 cites W2152728141 @default.
- W4280557595 cites W2510642588 @default.
- W4280557595 cites W2564013664 @default.
- W4280557595 cites W2734774145 @default.
- W4280557595 cites W2952218014 @default.
- W4280557595 cites W2962780374 @default.
- W4280557595 cites W2962866211 @default.
- W4280557595 cites W2972541922 @default.
- W4280557595 cites W2973054567 @default.
- W4280557595 cites W2973143779 @default.
- W4280557595 cites W3086154751 @default.
- W4280557595 cites W3095166612 @default.
- W4280557595 cites W3096893582 @default.
- W4280557595 cites W3150964372 @default.
- W4280557595 doi "https://doi.org/10.3389/frsip.2022.856968" @default.
- W4280557595 hasPublicationYear "2022" @default.
- W4280557595 type Work @default.
- W4280557595 citedByCount "3" @default.
- W4280557595 countsByYear W42805575952022 @default.
- W4280557595 countsByYear W42805575952023 @default.
- W4280557595 crossrefType "journal-article" @default.
- W4280557595 hasAuthorship W4280557595A5004802529 @default.
- W4280557595 hasAuthorship W4280557595A5027797344 @default.
- W4280557595 hasAuthorship W4280557595A5030528300 @default.
- W4280557595 hasBestOaLocation W42805575951 @default.
- W4280557595 hasConcept C111472728 @default.
- W4280557595 hasConcept C111919701 @default.
- W4280557595 hasConcept C115961682 @default.
- W4280557595 hasConcept C118505674 @default.
- W4280557595 hasConcept C138885662 @default.
- W4280557595 hasConcept C154945302 @default.
- W4280557595 hasConcept C163294075 @default.
- W4280557595 hasConcept C204201278 @default.
- W4280557595 hasConcept C2776182073 @default.
- W4280557595 hasConcept C2776864781 @default.
- W4280557595 hasConcept C28490314 @default.
- W4280557595 hasConcept C41008148 @default.
- W4280557595 hasConcept C60048801 @default.
- W4280557595 hasConcept C61328038 @default.
- W4280557595 hasConcept C99498987 @default.
- W4280557595 hasConceptScore W4280557595C111472728 @default.
- W4280557595 hasConceptScore W4280557595C111919701 @default.
- W4280557595 hasConceptScore W4280557595C115961682 @default.
- W4280557595 hasConceptScore W4280557595C118505674 @default.
- W4280557595 hasConceptScore W4280557595C138885662 @default.
- W4280557595 hasConceptScore W4280557595C154945302 @default.
- W4280557595 hasConceptScore W4280557595C163294075 @default.
- W4280557595 hasConceptScore W4280557595C204201278 @default.
- W4280557595 hasConceptScore W4280557595C2776182073 @default.
- W4280557595 hasConceptScore W4280557595C2776864781 @default.
- W4280557595 hasConceptScore W4280557595C28490314 @default.
- W4280557595 hasConceptScore W4280557595C41008148 @default.
- W4280557595 hasConceptScore W4280557595C60048801 @default.
- W4280557595 hasConceptScore W4280557595C61328038 @default.
- W4280557595 hasConceptScore W4280557595C99498987 @default.
- W4280557595 hasFunder F4320314731 @default.
- W4280557595 hasLocation W42805575951 @default.
- W4280557595 hasLocation W42805575952 @default.
- W4280557595 hasOpenAccess W4280557595 @default.
- W4280557595 hasPrimaryLocation W42805575951 @default.
- W4280557595 hasRelatedWork W1986772939 @default.
- W4280557595 hasRelatedWork W1998521395 @default.
- W4280557595 hasRelatedWork W2037635165 @default.
- W4280557595 hasRelatedWork W2046186789 @default.
- W4280557595 hasRelatedWork W2118159520 @default.
- W4280557595 hasRelatedWork W2140410589 @default.
- W4280557595 hasRelatedWork W2144314030 @default.
- W4280557595 hasRelatedWork W4210266374 @default.
- W4280557595 hasRelatedWork W4221152531 @default.
- W4280557595 hasRelatedWork W4375869276 @default.
- W4280557595 hasVolume "2" @default.
- W4280557595 isParatext "false" @default.
- W4280557595 isRetracted "false" @default.
- W4280557595 workType "article" @default.