Table 12
Statistics about the reduced corpora computed by ASA and LamSCP from Learning corpus. The last column concerns the best lower bound found by LamSCP.
| Speech synthesis experiment: corpus reduction | |||
|---|---|---|---|
| ASA | LamSCP |  | |
| Covering size (phones) | 36,538 | 35,681 | 35,655 |
| Reduction rate relative to the full corpus (%) | –90.70 | –90.92 | –90.92 |
| Reduction rate relative to ASA (%) | –2.35 | –2.42 | |
| Number of syntagms | 2,191 | 2,119 | – |
| Speech synthesis experiment: corpus reduction | |||
|---|---|---|---|
| ASA | LamSCP | | |
| Covering size (phones) | 36,538 | 35,681 | 35,655 |
| Reduction rate relative to the full corpus (%) | –90.70 | –90.92 | –90.92 |
| Reduction rate relative to ASA (%) | –2.35 | –2.42 | |
| Number of syntagms | 2,191 | 2,119 | – |