Samples
MUNRA builds a library of unique audio samples recorded specifically for training generative models. All samples are also available for direct use in music production.
Recording Philosophy
Unlike commercial sample libraries that aim for "perfection," these samples embrace imperfection: room tone, analog character, timing inconsistencies.
Every sample is:
- Recorded by hand
- Documented with metadata
- Used to train our models
- Free to use with attribution
Download
Coming Soon — First pack releasing March 2026
Current Library
Status: In progress (target: 2,000+)
License: CC-BY 4.0
Format: WAV, 48kHz, 24-bit
Categories
Drums (coming soon)
Voice (coming soon)
Synths (coming soon)