Grapheme-to-Phoneme and Prosody (G2PnP) conversion on 6D-Eval
1.71CER (PnP)Dict-DNN-NS
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Dict-DNN-NSType=Non-streaming, P (past context size)=∞, C (chunk size)=∞, M (minimum look-ahead size)=∞2026.02 | 1.71 | 40.4 | 1.18 | 26.4 | 0.4 | 6.4 | — | |
| CC-G2PnP-5-1Type=Streaming, P (past context size)=10, C (chunk size)=5, M (minimum look-ahead size)=12026.02 | 1.79 | 41.4 | 1.28 | 28.6 | 0.52 | 8.4 | — | |
| CC-G2PnP-5-2Type=Streaming, P (past context size)=10, C (chunk size)=5, M (minimum look-ahead size)=22026.02 | 1.8 | 41.4 | 1.3 | 28.9 | 0.53 | 8.6 | — | |
| CC-G2PnP-NSType=Non-streaming, P (past context size)=∞, C (chunk size)=∞, M (minimum look-ahead size)=∞2026.02 | 1.8 | 42 | 1.33 | 29.9 | 0.48 | 7.6 | — | |
| CC-G2PnP-2-2Type=Streaming, P (past context size)=10, C (chunk size)=2, M (minimum look-ahead size)=22026.02 | 1.85 | 42.4 | 1.33 | 29.8 | 0.53 | 8.7 | — | |
| CC-G2PnP-2-1Type=Streaming, P (past context size)=10, C (chunk size)=2, M (minimum look-ahead size)=12026.02 | 1.9 | 43.4 | 1.36 | 30 | 0.55 | 9 | — | |
| CC-G2PnP-5-0Type=Streaming, P (past context size)=10, C (chunk size)=5, M (minimum look-ahead size)=02026.02 | 2.01 | 45.3 | 1.45 | 33.4 | 0.62 | 11.3 | — | |
| Dict-DNN-20Type=Streaming, P (past context size)=0, C (chunk size)=20, M (minimum look-ahead size)=02026.02 | 2.28 | 45.9 | 1.72 | 36 | 0.56 | 8.7 | — | |
| CC-G2PnP-2-0Type=Streaming, P (past context size)=10, C (chunk size)=2, M (minimum look-ahead size)=02026.02 | 2.44 | 51.6 | 1.77 | 39.7 | 0.82 | 16.2 | — | |
| Dict-DNN-10Type=Streaming, P (past context size)=0, C (chunk size)=10, M (minimum look-ahead size)=02026.02 | 3.58 | 62.2 | 2.97 | 57.2 | 0.86 | 13 | — | |
| Dict-DNN-5Type=Streaming, P (past context size)=0, C (chunk size)=5, M (minimum look-ahead size)=02026.02 | 6.67 | 84.7 | 5.9 | 82.8 | 1.54 | 22.4 | — |