VCTK
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
VCTK (unseen speakers)
3.79MOS
7
VCTK (test)
2.1LSD
7
VCTK (Noisy-distorted)
1.16LSD (8→16kHz)
6
VCTK 16 kHz -> 48 kHz v0.92 (test)
0.705LSD
6
VCTK 4 kHz -> 48 kHz v0.92 (test)
0.877LSD
6
VCTK 48 kHz (test)
3.62ViSQOL
6
VCTK
1.5WER
6
VCTK 8 kHz input (test)
1.014LSD
6
VCTK 16 kHz input (test)
0.813LSD
6
VCTK GSR
3.128DNSMOS
6
VCTK-BWE BW=1K (test)
4.154WVMOS
6
VCTK zero-shot
1.9WER
6
VCTK
2.64PESQ
6
VCTK 4 kHz -> 16 kHz (test)
3.78WB-PESQ
6
VCTK 8 kHz -> 16 kHz (test)
4.28WB-PESQ
6
VCTK 4-16 kHz
0.94LSD
6
VCTK 8-16 kHz
0.77LSD
6
VCTK seen-to-seen (test)
4.32MOS
6
VCTK (Mixed)
77Accuracy
6
VCTK 48 kHz, 1.5 kbps (test)
3.305SIGMOS
5
VCTK 16-bit
2.68Compression Rate
5
VCTK (test)
0.74LSD
5
VCTK (test)
0.8LSD
5
VCTK (test)
0.85LSD
5
VCTK 16 kHz (test)
23.58SNR
5