Source code: https://github.com/hash2430/pitchtron/
Audio samples
| Prosody | Reference | GST | Soft pitchtron | Hard pitchtron |
|---|---|---|---|---|
| Standard Korean neutral dialogue | ||||
| Standard Korean neutral dialogue | ||||
| Standard Korean emotive dialogue | ||||
| Standard Korean emotive dialogue | ||||
| Kyoungsang dialect | ||||
| Kyoungsang dialect | ||||
| Cheolla dialect | ||||
| Cheolla dialect |
Prosody transferability and vocal range scalability
| Scale | Reference | GST | Soft pitchtron | Hard pitchtron |
|---|---|---|---|---|
| As is | ||||
| Scale pitch by 0.5 | - |
Non-linear prosody transfer
| Reference | GST | Soft pitchtron |
|---|---|---|
'TTS' 카테고리의 다른 글
| FastSpeech: Fast, Robust and Controllable Text to Speech (0) | 2020.05.28 |
|---|---|
| CHiVE: Varying prosody in speech synthesis with a linguistically driven dynamic hierarchical conditional variational network (0) | 2020.05.25 |
| Expressive TTS and prosody transfer (0) | 2020.04.20 |
| E2E TTS에서 postnet이 필요한 이유 (0) | 2020.03.14 |
| Japanese/Korean/Vietnamese Corpus (0) | 2020.02.08 |
