Demo index

Audio Source Separation Using Multiple Deformed References

by N. Souviraà-Labastie, A. Olivero, E. Vincent, F. Bimbot


Note: The results of the -6 dB (resp. 12 dB) case are inferior (resp. superior) to the average reported in the Table 1 of the paper.

Voice-to-music ratio levels : -6dB . Music as foreground and voice as background
Voice Music
Groundtruth

Mixture

References
Separated using speech reference
Separated using music reference
Separated using speech and music references

Voice-to-music ratio levels : 12dB . Voice as foreground and music as background
Voice Music
Groundtruth

Mixture

References
Separated using speech reference
Separated using music reference
Separated using speech and music references