VCTK
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
VCTK In-domain
0.6LSD
34
VCTK (test)
1.01Log Spectral Distance
28
VCTK
0WER
27
VCTK++ (test)
93F1 Score
25
VCTK noisy
1.52WER
24
VCTK
1.7WER
19
VCTK 48 kHz (test)
0.895STOI
18
VCTK Accelerometer 12-bit, 4-16 kHz upsampling (test)
0.87LSD
18
VCTK Vibration sensor 12-bit, 4-16 kHz upsampling (test)
0.84LSD (Log-Spectral Distance)
18
VCTK 0.92 (test)
0.7LSD
16
VCTK English
4.53NISQA-MOS
15
VCTK (test)
3.47WER
15
VCTK Multi-speaker (test)
20SNR
15
VCTK Single-speaker (test)
19.5SNR
15
VCTK A→T
96.1Recall@1
15
VCTK (10% unseen utterances)
4.05MOS
15
VCTK (10% unseen utterances)
3.98MOS
15
VCTK en
2.476PESQ
14
VCTK 10H budget v0.92 (train)
9.9WER
12
VCTK 5H training budget v0.92
8.5WER (%)
12
VCTK 1H training budget v0.92
7.5WER
12
VCTK 30m budget v0.92 (train)
10WER
12
VCTK 10m budget v0.92 (train)
42.2WER
12
VCTK
-35.65BD-Rate ViSQOL
12
VCTK 48 kHz (test)
0.142RTF (CPU)
12