Soundprism Examples

This is the companion webpage for the paper:

Zhiyao Duan and Bryan Pardo, Soundprism: an online system for score-informed source separation of music audio, IEEE Journal of Selected Topics in Signal Processing, vol. 5, no. 6, pp. 1205-1215, 2011. <pdf> <slides>

Source separation examples

As describied in the paper, we compare Soundprism with three other source separation systems:

1. We first present three examples from the Bach chorale dataset, each of which has a different polyphony.

ID Mixture Sources Soundprism Ideally-aligned Ganseman10 MPET
1 MIDI auido violin violin violin violin violin
clarinet clarinet clarinet clarinet clarinet
2 MIDI auido clarinet clarinet clarinet clarinet clarinet
saxophone saxophone saxophone saxophone saxophone
bassoon bassoon bassoon bassoon bassoon
3 MIDI auido violin violin violin violin violin
clarinet clarinet clarinet clarinet clarinet
saxophone saxophone saxophone saxophone saxophone
bassoon bassoon bassoon bassoon bassoon

2. Then, we present two examples of realistic orchestral music from the RWC database [4], where sources are mixed in a natural environment instead of being individually recorded then artificially mixed. Note that we do not have the ground-truth sources, nor the ground-truth audio-score alignment. Therefore, we do not have results of "Ideally-aligned" here.

ID Mixture Soundprism Ganseman10 MPET
4 MIDI auido clarinet clarinet clarinet
viola viola viola
cello cello cello
5 MIDI auido violin1 violin1 violin1
violin2 violin2 violin2
viola viola viola
cello cello cello

Score following results

Now we show score following results of the above examples. We compare Soundprism with an offline audio-score alignment method:

We show the results by time-warping the original mixture according to the output alignment of each score follower/aligner. You may follow the following steps to get a feel for the alignment results of these examples:

  1. Import the score file (*.mid) to Audacity;
  2. Import the time-warped audio file (*.wav) to Audacity;
  3. Listen to the audio and watch how well it is aligned with the score.
ID Score Original audio Time-warped audio by Soundprism Time-warped audio by Scorealign
1 MIDI wave wave wave
2 MIDI wave wave wave
3 MIDI wave wave wave
4 MIDI wave wave wave
5 MIDI wave wave wave

References

[1] J. Ganseman, G. Mysore, P. Scheunders and J. Abel, "Source separation by score synthesis," in Proc. International Computer Music Conference (ICMC), New York, NY, June 2010.
[2] Z. Duan, B. Pardo and C. Zhang, "Multiple fundamental frequency estimation by modeling spectral peaks and non-peak regions," IEEE Trans. Audio Speech Language Process., vol. 18, no. 8, pp. 2121-2133, 2010.
[3] Z. Duan, J. Han and B. Pardo, "Song-level multi-pitch tracking by heavily constrained clustering," in Proc. IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), 2010, pp. 57-60.
[4] M. Goto, H. Hashiguchi, T. Nishimura and R. Oka, "RWC music database: popular, classical, and jazz music databases," in Proc. International Conference on Music Information Retrieval (ISMIR 2002), 2002, pp.287-288.
[5] N. Hu, R.B. Dannenberg and G. Tzanetakis, "Polyphonic audio matching and alignment for music retrieval," in Proc. 2003 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), New Paltz, New York, USA, 2003, pp. 185-188.

For any questions or comments, please contact us at zhiyaoduan00 AT gmail DOT com.