All snippets generated with musicgen, a generative model to create music from text. The model has been created by Facebook AI Research (FAIR) and was released on huggingface, huggingface.co/spaces/facebook/MusicGen
The snippets are meant for gapless playback, which means there should not be any pause between the tracks. Streaming from bandcamp cannot do this, so you need to download the files to your computer and use an audio player for gapless playback locally, on Linux audacious works, on MacOS maybe VLC, on Windows we record that foobar2000 is good, winamp works. Send your solutions.
For additional convenience we provide all files spliced together into one single track. It is the first track of the album. The benefit is, this file also includes the tracks with emojis in their filenames, which are currently not supported to be uploaded by bandcamp.
The text prompts that were used to generate the snippets are recorded directly in the track titles.
The resulting set of snippets was sorted according to a distance function based on a Laion-CLAP encoding to give the final tracklist. This sorting packs similar sounding snippets into contiguous blocks, so they form mini-tracks within the overall composition.
No additional processing was applied, no autoedit, no mastering.
prompt generation and sorting by farmersmanual
time from idea to release < 24 hours
unmastered
farmersmxnual presents - generate and test - gt101