tracert/ukr-tts
1
Still in the progress.
Before training, the dataset "Tetiana" was cleared of noise and low-frequency hum. Blowing into the microphone at medium and low frequencies was also filtered. The model size is only 150 MB.
- Thank Yehor Smoliakov for dataset Tetiana Ukrainian dataset - Yehor Smoliakov @egorsmkv
- Tetiana (female): Original dataset example