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

Non-parallel voice conversion (VC) is a technique for training voiceconverters without a parallel corpus. Cycle-consistent adversarialnetwork-based VCs (CycleGAN-VC and CycleGAN-VC2) are widely accepted asbenchmark methods. However, owing to their insufficient ability to grasptime-frequency structures, their application is limited to mel-cepstrumconversion and not mel-spectrogram conversion despite recent advances inmel-spectrogram vocoders. To overcome this, CycleGAN-VC3, an improved variantof CycleGAN-VC2 that incorporates an additional module called time-frequencyadaptive normalization (TFAN), has been proposed. However, an increase in thenumber of learned parameters is imposed. As an alternative, we proposeMaskCycleGAN-VC, which is another extension of CycleGAN-VC2 and is trainedusing a novel auxiliary task called filling in frames (FIF). With FIF, we applya temporal mask to the input mel-spectrogram and encourage the converter tofill in missing frames based on surrounding frames. This task allows theconverter to learn time-frequency structures in a self-supervised manner andeliminates the need for an additional module such as TFAN. A subjectiveevaluation of the naturalness and speaker similarity showed thatMaskCycleGAN-VC outperformed both CycleGAN-VC2 and CycleGAN-VC3 with a modelsize similar to that of CycleGAN-VC2. Audio samples are available athttp://www.kecl.ntt.co.jp/people/kaneko.takuhiro/projects/maskcyclegan-vc/index.html.

MaskCycleGAN-VC: Learning Non-parallel Voice Conversion with Filling in Frames