VoIP Round 1 Results

Score was calculated as described in contest evaluation criteria: sum(max(|x_i - y_i| - 0.3, 0.0)^2). Lesser value is better.

Entry Score Notes
fake_correct 16 Minimum possible score
997 335 4.75⭐️, 1st place
1012 551 3.8⭐️, 2nd place
1007 649 3.3⭐️, 2nd place
1010 653 4⭐️, 2nd place
fake_correct_1v5 731 1 was returned if user_score < 3, 5 otherwise
1002 795 2.75⭐️, 3rd place
994 828 2.7⭐️, 3rd place
1013 832 3⭐️, 3rd place
987 839 2.75⭐️, 3rd place
1009 919 2.75⭐️, 3rd place
988 956 2.85⭐️, 3rd place
995 954 2.5⭐️, 3rd place
1006 1 063 2⭐️, 4th place
fake_always_3 1 228 3 is always returned as result
986 1 228
993 1 359
1004 1 781
1014 1 838
1005 2 096
1000 2 280
984 2 504
1011 3 121
1001 3 141
fake_always_1 3 645 1 is always returned as result
996 3 683
998 4 822
fake_always_5 4 973 5 is always returned as result
992 6 360

Dataset

Distorted audios

Download: distorted.zip

Each filename looks like:
entry{call_entry}_{original}.ogg_{network_conditions}.dur.ogg

call_entry - Contest entry number of tgvoipcall, which was used to produce distorted ogg.
original_sample_name - Original sample filename from VoIP_Round1_Test_Sounds.zip, 6.49MB
network_conditions - Comma separated list of network conditions.

Ratings provided by users and contestants

Download: ratings.csv

Columns:
original - Original sample filename from VoIP_Round1_Test_Sounds.zip, 6.49MB
call_entry - Contest entry number of tgvoipcall, which was used to produce distorted ogg.
net - Network conditions
correct_mean - Mean user rating for current pair
correct_cnt - Number of user ratings
fake_correct,fake_correct_1v5,fake_always_3,fake_always_1,fake_always_5 - Theoretical entries ratings, just for reference and comparison
984-1014 - Contestant entries ratings

Each line corresponds to 1 file from distorted.zip