We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
John Sandall - Building A Folk Music Recommendation System with LLMs | PyData London 2024
Build a folk music recommendation system using Large Language Models (LLMs) and explore novel applications in music composition, analysis, and prediction.
- Use LLMs to build a folk music recommendation system
- Concatenate folk tunes to create a large dataset, which is not normally distributed
- Apply Gaussian mixture models and topic modeling using embeddings to analyze the data
- Use clustering techniques, such as agglomerative clustering and bisecting k-means, to group similar tunes together
- Apply transformers to generate new folk tunes and assist in music composition
- Use ABC notation to input data and generate embeddings for analysis
- Explore the use of dimensionality reduction techniques, such as UMAP, to visualize the data
- Use cosine similarity to measure the similarity between tunes
- Apply gradient boost regressor models to predict the popularity of folk tunes
- Use TF-IDF and LDA topic modeling to analyze lyrical content and discover topics in folk music
- Utilize prompt engineering to improve the performance of the system