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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The diffusion model is capable of generating high-quality data through aprobabilistic approach. However, it suffers from the drawback of slowgeneration speed due to the requirement of a large number of time steps. Toaddress this limitation, recent models such as denoising diffusion implicitmodels (DDIM) focus on generating samples without directly modeling theprobability distribution, while models like denoising diffusion generativeadversarial networks (GAN) combine diffusion processes with GANs. In the fieldof speech synthesis, a recent diffusion speech synthesis model calledDiffGAN-TTS, utilizing the structure of GANs, has been introduced anddemonstrates superior performance in both speech quality and generation speed.In this paper, to further enhance the performance of DiffGAN-TTS, we propose aspeech synthesis model with two discriminators: a diffusion discriminator forlearning the distribution of the reverse process and a spectrogramdiscriminator for learning the distribution of the generated data. Objectivemetrics such as structural similarity index measure (SSIM), mel-cepstraldistortion (MCD), F0 root mean squared error (F0 RMSE), short-time objectiveintelligibility (STOI), perceptual evaluation of speech quality (PESQ), as wellas subjective metrics like mean opinion score (MOS), are used to evaluate theperformance of the proposed model. The evaluation results show that theproposed model outperforms recent state-of-the-art models such as FastSpeech2and DiffGAN-TTS in various metrics. Our implementation and audio samples arelocated on GitHub.

Adversarial Training of Denoising Diffusion Model Using Dual Discriminators for High-Fidelity Multi-Speaker TTS