en
Hot
en
Hot
en
Submit a Listing
Login/Register
#Training
diffae
This project introduces diffusion autoencoders that focus on meaningful and decodable image representation. Featured in CVPR 2022, it offers practical tools like Colab walkthroughs and web demos for sample generation, manipulation, and interpolation. The comprehensive documentation and LMDB datasets support ease of use. It also provides training and evaluation scripts for datasets such as FFHQ and CelebAHQ, facilitating advancements in AI image processing, and supplying essential tools for researchers and developers.
BigVGAN
BigVGAN presents a universal neural vocoder that refines speech synthesis by undergoing extensive training on varied audio datasets. It features rapid inference achieved through custom CUDA kernels and allows up to 44 kHz sampling rate for superior audio outcomes. Utilizing advanced multi-scale sub-band CQT discriminators and multi-scale mel spectrogram loss, it enhances audio fidelity and minimizes perceptual distortions, making it an essential asset for professionals in audio processing and synthesis.
fontogen
Explore how FontoGen facilitates font creation using AI and open-source datasets. The guide provides clear installation and training instructions for generating fonts. Ideal for both professionals and hobbyists, it outlines the process of training models on OFL fonts and invites contributions for model improvement.
Terms of Use
Privacy Policy
Advertising Services
Feedback Email: service@vectorlightyear.com