whisper-base-int8-ov

OpenVINO IR export of openai/whisper-base, quantized to INT8 — 79 MB.

Quantization weight compression (optimum)
Lesson 06 Whisper speech recognition
Built by convert/convert_all.py of the ARCademy OpenVINO courseware

Load with openvino_genai.WhisperPipeline(model_dir, device); feed 16 kHz mono float samples.

from huggingface_hub import snapshot_download
model_dir = snapshot_download("circulus/whisper-base-int8-ov")
Downloads last month
30
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for circulus/whisper-base-int8-ov

Finetuned
(762)
this model