Free Information Center
MusicGen is a text-to-music generation model developed by Meta that produces musical audio from textual prompts. It leverages deep learning techniques to generate diverse and coherent musical compositions based on user-provided descriptions.
AudioLDM is a machine learning framework that applies latent diffusion models to audio generation and processing. It leverages latent space representations to efficiently synthesize high-quality audio from text or other audio inputs.
Variational autoencoders (VAEs) are generative models that leverage neural networks to learn complex data distributions, enabling effective data generation and representation.
A flow-based generative model is a type of machine learning model that generates data by learning invertible mappings between complex data distributions and simple latent variables. These models enable exact likelihood computation and efficient sampling, distinguishing them from other generative approaches.
A variational diffusion model is a type of generative model that combines principles from variational inference and diffusion processes to generate data through a controlled stochastic process. It is used primarily in machine learning to model complex data distributions by gradually transforming noise into structured data.
RealNVP (real-valued non-volume preserving) is a type of generative model in machine learning that utilizes invertible neural networks for density estimation and sampling. It enables exact log-likelihood computation and efficient generation of complex data distributions, particularly in image modeling.
VQ-VAE (Vector Quantized Variational Autoencoder) is a generative model architecture that combines discrete latent representations with variational autoencoders, enabling efficient learning and synthesis of complex data like images and audio.
GET3D is a generative 3D model designed to produce high-quality three-dimensional objects from 2D images. It leverages advances in machine learning to generate detailed 3D shapes and textures, enabling applications in graphics, gaming, and virtual reality.
The Wake–sleep algorithm is a computational method used in machine learning to train probabilistic generative models, particularly useful in unsupervised learning. It alternates between wake and sleep phases to refine model parameters and recognition networks.
Diffsound (discrete diffusion for audio) is a method in generative audio modeling that applies discrete diffusion processes to create or transform audio data. It is part of a broader class of diffusion-based generative models adapted specifically for the discrete, sequential nature of audio signals.