OpenAI Jukebox Description
Jukebox is a neural net that generates music. It can also rudimentary sing. We are releasing the code and model weights, as well as a tool to help you explore the generated samples. Jukebox generates a new music sample from scratch when you provide genre, artist, lyrics, and other inputs. Jukebox can produce a wide variety of music and singing styles, and it can also generalize to lyrics that were not present during training. All of the below lyrics were co-written by OpenAI researchers and a language model. Jukebox creates songs that are very different from the original songs when it is trained using lyrics learned during training. Jukebox provides 12 seconds of audio for conditioning and then completes the rest in a specific style. We chose music to work on because we want to push the boundaries of generative modeling. Jukebox's autoencoder compresses audio into a discrete space using a quantization-based method called VQ-VAE.
Pricing
Company Details
Product Details
OpenAI Jukebox Features and Options
OpenAI Jukebox User Reviews
Write a Review- Previous
- Next