# DEMO + Results
# Demo Tracks
# Objective results on SiSEC 2018
Open-Unmix yields state-of-the-art results compared to participants from SiSEC 2018. The performance of
UMX is almost identical since it was evaluated on compressed STEMS.
It can be seen that our proposed model reaches state-of-the-art results. There is no
statistically significant difference between the best method
TAK1 is not released as open-source, this indicates
that Open-Unmix is the current state-of-the-art open-source source
UMXHQ] were omitted as were trained on only MUSDB18.
note that the scores represent (Median of frames, Median of tracks)
If you want to compare the results to of
umxhq to your own methods, the evaluation scores are available for download here: