GitHub - samim23/polymath: Convert any music library into a music production sample-library with ML
πΆπ€ Dive into the world of music production like never before with Polymath! π΅π₯ This AI tool uses machine learning to convert any music library into a sample-library, separate stems, analyze musical elements, and more. Perfect for DJs, composers, and ML developers. π§πΉ #AI #MusicProduction #ML #Polymath
- **Project Name:** Polymath
- **Description:** Converts any music library into a music production sample-library using machine learning.
- **Features:** Separates songs into stems, quantizes to the same tempo, analyzes musical structure, key, timbre, loudness, etc., and converts audio to MIDI.
- **Use Cases:** Facilitates creating new compositions by combining elements from various songs effortlessly. Enables creating mash-up DJ sets quickly. Simplifies creating large music datasets for ML developers.
- **Technical Details:** Utilizes neural networks like Demucs, sf_segmenter, Crepe, pyrubberband, librosa for tasks like source separation, structure segmentation, pitch tracking, key detection, MIDI transcription, quantization, alignment, and info retrieval.
- **Requirements:** Python >=3.7 and <=3.10, ffmpeg, GPU support through cuda for native setups.
- **Installation:** Clone the repository and install requirements using pip.
- **Docker Support:** Dockerfile provided for easy setup; allows input/output file exchange between host system and container.
- **Usage:** Add songs to the library, quantize songs, search for similar songs, and convert audio to MIDI using specific commands.
- **License:** Released under the MIT license.
- **Community:** Join the Polymath community on Discord.