In this video I'm focusing on a technique to force RAVE models' to generate loops.
The decoder/ generator unit of a RAVE model translates latent encodings into audio signals. Latent encodings are basically multidimensional vectors of signal values which in an Autoencoder architecture are created by the encoder's translation of an input signal.
In the patch shown in this video the latent encodings are emulated with the noise~ object in Pure Data, which generates pseudo random signal values between -1 and 1. By feeding pre-defined seeds to noise~, the sequence of this signal values is reproducible. With *~ the magnitude of signal values can be altered; sig~ offsets its center value.
RAVE is "A variational autoencoder for fast and high-quality neural audio synthesis” created by Antoine Caillon and Philippe Esling of Artificial Creative Intelligence and Data Science (ACIDS) at IRCAM, Paris.
RAVE on GitHub: github.com/aci...
nn~ on GitHub: github.com/aci...
To train RAVE models on Colab or Kaggle, you can use these Jupyter notebooks i've set up: github.com/dev...
18 сен 2024