Neural Vocoding on LJSpeech 88 (test)
0.9M-STFTBigVGAN
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| BigVGAN#Param. (M)=112.18, #MACs (Giga/5s)=417.20, Inference Speed CPU=0.6674(1.50×), Inference Speed GPU=0.0507(20×)2026.03 | 0.9 | 4.107 | 1.877 | 0.084 | 97.2 | 20.692 | 4.314 | 4.87 | |
| RNDVoC-shared#Param. (M)=3.14, #MACs (Giga/5s)=34.10, Inference Speed CPU=0.0669(14.96×), Inference Speed GPU=0.0043(233×)2026.03 | 0.91 | 3.987 | 2.047 | 0.085 | 97.1 | 21.318 | 4.161 | 4.837 | |
| HiFTNet#Param. (M)=22.43, #MACs (Giga/5s)=163.05, Inference Speed CPU=0.1559(6.41×), Inference Speed GPU=0.0085(117×)2026.03 | 0.923 | 3.992 | 2.039 | 0.092 | 96.6 | 20.78 | 4.343 | 4.849 | |
| RNDVoC-nonshared#Param. (M)=9.48, #MACs (Giga/5s)=24.98, Inference Speed CPU=0.0770(12.98×), Inference Speed GPU=0.0063(158×)2026.03 | 0.934 | 3.892 | 2.214 | 0.089 | 97 | 21.952 | 4.138 | 4.812 | |
| BigVGAN-base#Param. (M)=13.94, #MACs (Giga/5s)=152.90, Inference Speed CPU=0.3856(2.59×), Inference Speed GPU=0.0368(27×)2026.03 | 0.978 | 3.603 | 2.331 | 0.12 | 95.6 | 30.277 | 4.208 | 4.822 | |
| APNet2#Param. (M)=31.38, #MACs (Giga/5s)=13.53, Inference Speed CPU=0.0187(53.48×), Inference Speed GPU=0.0016(643×)2026.03 | 0.982 | 3.492 | 2.829 | 0.113 | 95.9 | 25.363 | 3.984 | 4.752 | |
| FreeV#Param. (M)=18.19, #MACs (Giga/5s)=7.84, Inference Speed CPU=0.0139(71.94×), Inference Speed GPU=0.0011(951×)2026.03 | 1.008 | 3.593 | 2.75 | 0.112 | 96 | 25.992 | 4.01 | 4.743 | |
| Vocos#Param. (M)=13.46, #MACs (Giga/5s)=5.80, Inference Speed CPU=0.0066(151.52×), Inference Speed GPU=0.0007(1400×)2026.03 | 1.012 | 3.522 | 2.67 | 0.121 | 95.6 | 29.13 | 3.968 | 4.774 | |
| Avocodo#Param. (M)=13.94, #MACs (Giga/5s)=152.95, Inference Speed CPU=0.1611(6.21×), Inference Speed GPU=0.0063(158×)2026.03 | 1.165 | 3.604 | 3.657 | 0.139 | 94.6 | 32.989 | 4.152 | 4.771 | |
| HiFiGAN-V1#Param. (M)=13.94, #MACs (Giga/5s)=152.90, Inference Speed CPU=0.1669(5.99×), Inference Speed GPU=0.0069(145×)2026.03 | 1.167 | 3.574 | 3.671 | 0.134 | 94.7 | 33.687 | 4.218 | 4.771 | |
| iSTFTNet-V1#Param. (M)=13.26, #MACs (Giga/5s)=107.76, Inference Speed CPU=0.0949(10.54×), Inference Speed GPU=0.0038(266×)2026.03 | 1.188 | 3.535 | 3.625 | 0.135 | 94.7 | 35.106 | 4.236 | 4.756 | |
| APNet#Param. (M)=72.19, #MACs (Giga/5s)=31.11, Inference Speed CPU=0.0300(33.33×), Inference Speed GPU=0.0022(458×)2026.03 | 1.266 | 3.39 | 3.285 | 0.151 | 94.5 | 23.057 | 3.17 | 4.695 |