snakers4/silero-stress
Silero Stress — pre-trained enterprise-grade automated stress and homograph disambiguation for the Russian language
Provides independent stress placement and homograph disambiguation modules covering ~4M Russian word forms with 100% accuracy on known words and 85 F1-score on ~2K homographs. Built on PyTorch with minimal dependencies (50MB package), achieves ~0.5ms per-word latency on single-threaded CPU and generalizes to unknown words at 60-70% accuracy. Designed for text-to-speech, telephony, and voice bot applications with zero telemetry or vendor lock-in.
125 stars and 1,341 monthly downloads. Available on PyPI.
Stars
125
Forks
3
Language
Python
License
MIT
Category
Last pushed
Feb 12, 2026
Monthly downloads
1,341
Commits (30d)
0
Dependencies
1
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/voice-ai/snakers4/silero-stress"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related tools
snakers4/silero-models
Silero Models: pre-trained text-to-speech models made embarrassingly simple
JSchmie/ScrAIbe-WebUI
WebUI for ScAIbe
abus-aikorea/voice-pro
Gradio WebUI for creators and developers, featuring key TTS (Edge-TTS, kokoro) and zero-shot...
isaiahbjork/orpheus-tts-local
Run Orpheus 3B Locally With LM Studio
ycyy/edge-tts-webui
edge-tts webui