trained from 1300 sentences
using 10 sentences per speaker
trained from 1300 sentences
using 10 sentences per speaker
Mirjam Wester and Reima Karhila:
Speaker similarity evaluation of foreign-accented speech synthesis using HMM-based speaker adaptation
Speaker adaptation with 10 sentences, as used in the listening tests for ICASSP 2011.
(See the paper in IEEExplore)
| Original target speaker sample (in English) | |
Speaker 1: |
Speaker 2: |
||
| Average American English voice trained from 1300 sentences | |
||||
| Adapted American English voice using 10 sentences per speaker | |
Speaker 1: |
Speaker 2: |
||
| Average Finnish English voice trained from 1300 sentences | | ||||
| Adapted Finnish English voices using 10 sentences per speaker | |
Speaker 1: |
Speaker 2: |
||