Layered Speech Infrastructure
Production-Ready TTS Infrastructure
Transform open-source TTS models into streaming-ready systems. We harden decoding, orchestration, and delivery so you can focus on the product moments that matter.
Indic Section
Indic flow · Input → Generate → Model
Choose from Indic-19 (streaming), Indic F5 (high-quality), or Chatterbox Multilingual (Hindi only).
Indic Input
Optimized for Indic models
Indic Generate
Streams Indic-19 with live PCM playback.
Indic-19 Direct
Modal proxy > Sub200 Indic stream
IndicBest for Hindi and Indic languages; tuned voices handle multilingual input.
Curated # / ## Indic voices tuned for low-latency streaming.
Each prompt POSTs with stream: false; full audio is generated before playback is enabled.
Advanced controls
Only sent if explicitly set. Default is 24000 Hz (not sent as parameter). Use 8000 / 16000 for lighter payloads.
When normalization is enabled, specify language code for better accuracy. Leave as "Autodetect" to let the API detect the language automatically.
Triggered by the global Generate button.