English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We propose a novel speech separation model designed to separate mixtures withan unknown number of speakers. The proposed model stacks 1) a dual-pathprocessing block that can model spectro-temporal patterns, 2) a transformerdecoder-based attractor (TDA) calculation module that can deal with an unknownnumber of speakers, and 3) triple-path processing blocks that can modelinter-speaker relations. Given a fixed, small set of learned speaker queriesand the mixture embedding produced by the dual-path blocks, TDA infers therelations of these queries and generates an attractor vector for each speaker.The estimated attractors are then combined with the mixture embedding byfeature-wise linear modulation conditioning, creating a speaker dimension. Themixture embedding, conditioned with speaker information produced by TDA, is fedto the final triple-path blocks, which augment the dual-path blocks with anadditional pathway dedicated to inter-speaker processing. The proposed approachoutperforms the previous best reported in the literature, achieving 24.0 and23.7 dB SI-SDR improvement (SI-SDRi) on WSJ0-2 and 3mix respectively, with asingle model trained to separate 2- and 3-speaker mixtures. The proposed modelalso exhibits strong performance and generalizability at counting sources andseparating mixtures with up to 5 speakers.

Boosting Unknown-number Speaker Separation with Transformer Decoder-based Attractor