An open multilingual TTS model for expressive, human-like speech across India's languages.
Visit svara-tts for more details.
Higher = more expressive prosody; 0.6-0.9 for conversational, 0.9-1.2 for dramatic
0.6-0.8 for natural prosody, 0.8-1.0 for expressive/dramatic
≥1.1 recommended for stable generation; prevents loops
Typical range: 900-1200 for most sentences