MSRBench: A Benchmarking Dataset for Music Source Restoration

MSRBench is the validation dataset for the MSR Challenge 2025 (https://msrchallenge.com); it contains 250 professionally mixed audio clips and their corresponding ground-truth targets under both original mix and 12 degradation scenarios in three categories.

Download

hf download yongyizang/MSRBench --repo-type dataset --local-dir {your output dir}

Total file size after unzipping is 28.7 GB.

Folder Structure

{root-of-this-repo}/
└── {Stem_Name}.zip
(after unzipping...)
    ├── mixture/
    │   ├── {song_id}_DT{degradation_type_id}.flac
    │   ├── {song_id}_DT{degradation_type_id}.flac
    │   └── ...
    └── targets/
        ├── {song_id}.flac
        ├── {song_id}.flac
        └── ...

Every *.flac file is 48 kHz stereo; each {stem_name} contains 250 song_ids, corresponding to 250 song clips; each song_id has 13 degradation types.

Available Stems

• Vocals
• Guitars
• Bass
• Keyboards
• Synthesizers
• Drums
• Percussions
• Orchestral Elements

Degradation Types

DT0: Original mixture produced by mixing engineers.

DT1-4: Analog and Environmental Distortions

Radio (DT1)

Simulates FM broadcasting using GNU Radio's stereo multiplex encoding/decoding blocks with standard broadcast parameters:

Signal Processing Parameters:

• Audio sampling rate: 44.1 kHz
• Quadrature rate: 220.5 kHz
• Pre-emphasis/de-emphasis time constant: 75 μs (US standard)
• Maximum frequency deviation: 75 kHz
• Stereo pilot tone: 19 kHz

Channel Model:

• Rayleigh fading: 8-sinusoid sum-of-sinusoids model with 20 Hz maximum Doppler shift
• Carrier frequency offset: σ = 0.1, max 1 kHz variations
• Multipath channel: 3-tap configuration
  • Delays: [0, 0.1, 0.5] ms
  • Relative magnitudes: [1.0, 0.8, 0.3]
• Noise: Additive white Gaussian noise at approximately 26 dB SNR (noise amplitude = 0.05)

Cassette (DT2)

Models magnetic tape coloration and noise using the DAW Cassette plugin by Klevgrand.[^1]

Vinyl (DT3)

Reproduces playback artifacts including crackle and mechanical noise using iZotope Vinyl,[^2] configured with the "1970" preset.

Live Sound (DT4)

Generated using:

• Impulse responses from PyRoomAcoustics
• Bandpass filtering to approximate phone microphone characteristics
• Environmental noise from the WHAM! dataset 48 kHz version mixed at approximately 20 dB SNR

DT5-8: Traditional Lossy Audio Codecs

DT5: 64 Kbps AAC

DT6: 64 Kbps MP3

DT7: 128 Kbps AAC

DT8: 128 Kbps MP3

DT9-12: Neural Audio Codecs

DT9: 22 kHz Descript Audio Codec (DAC)

DT10: 44 kHz DAC

DT11: 6 Kbps Encodec

DT12: 3 Kbps Encodec

---

[^1]: Klevgrand DAW Cassette [^2]: iZotope Vinyl