Unhallucinated Faster Whisper

Install the package

pip install unhallucinated-faster-whisper

Usage (Same as Faster Whisper)

from faster_whisper import WhisperModel 
model_size = "turbo"
model = WhisperModel(model_size, device="cpu")  

Enabling the Unhallucination Feature

Simply pass unhallucinated=True when transcribing

segments, info = model.transcribe("voice.mp3", unhallucinated=True)

Model Compatibility

Compatible with all Whisper models.