Piper TTS Nepali Text-to-Speech AI Training
Developed a Nepali text-to-speech AI model utilizing the Piper TTS toolkit and openSLR dataset. Conducted supervised model training from scratch to generate human-like audio output from Nepali text inputs. Task included curation and preparation of high-quality paired text and audio data for model training. • Focused on AI model performance improvement for language translation tasks. • Collected, cleaned, and prepared text and audio data for TTS training. • Iteratively evaluated generated speech for quality assurance. • Leveraged Piper TTS software and open-source resources throughout the pipeline.