MusiConGen
Text to Music Generation Model Based on Transformer
- Support text description to generate music samples
- Chord and rhythm control represented by symbols
- Generate music by combining multiple text description styles
- Estimate the chords of the generated sample through BTC chord recognition model
- Provide samples of different music styles (such as blues, jazz, rock, funk, heavy metal)
- Support comparing the performance of different fine-tuning methods
Product Details
MusiConGen is a Transformer based text to music generation model that enhances control over rhythm and chords through temporal conditions. This model is fine tuned from the pre trained MusicGen meta framework. It uses chord and rhythm controls represented by symbols, combined with five different styles of text descriptions to generate samples. The chords generated from the sample are estimated using the BTC chord recognition model, as described in the paper.
