phonikud and phonikud-tts
The phonikud-tts tool builds upon phonikud by using its grapheme-to-phoneme conversion as a prerequisite step in a complete Hebrew text-to-speech pipeline, making them complements rather than competitors.
About phonikud
thewh1teagle/phonikud
Hebrew grapheme to phoneme (G2P)
Uses an ONNX-based neural model to add diacritical marks (nikud) to unvocalized Hebrew text, then converts to IPA phonemes via rule-based finite-state transducers that handle modern Hebrew phonology including stress placement, Vocal Shva, and Hat'ama distinctions. Supports mixed Hebrew-English text with fallback phonemization and integrates with TTS systems, enabling accurate speech synthesis from plain text without manual annotation.
About phonikud-tts
thewh1teagle/phonikud-tts
phonikud-tts - text to speech in Hebrew
Implements neural text-to-speech for Hebrew using ONNX Runtime models with optional GPU acceleration via CUDA, building on the Phonikud phonetic engine. Supports CPU and GPU inference through configurable execution providers, enabling efficient speech synthesis across different hardware configurations. Available as a Python library with example implementations and a Hugging Face Space demo.
Scores updated daily from GitHub, PyPI, and npm data. How scores work