W

Whisper Large V2 Onnx Int4 Inc

Developed by Intel
Whisper is a pre-trained automatic speech recognition (ASR) and speech translation model, trained on 680,000 hours of labeled data, demonstrating strong generalization capabilities. This repository contains the INT4 weight-only quantized version of the Whisper large v2 model in ONNX format.
Downloads 19
Release Time : 10/8/2023

Model Overview

Whisper is a powerful automatic speech recognition and speech translation model that can adapt to various datasets and domains without fine-tuning. This model is an INT4 weight-only quantized version, powered by Intel® Neural Compressor.

Model Features

INT4 Weight-only Quantization
The model undergoes INT4 weight-only quantization, significantly reducing model size while maintaining high recognition accuracy.
Strong Generalization Capability
Trained on 680,000 hours of labeled data, the model can adapt to various datasets and domains without fine-tuning.
ONNX Format
The model is provided in ONNX format, facilitating deployment and inference across different platforms.

Model Capabilities

Automatic Speech Recognition
Speech Translation

Use Cases

Speech Recognition
Speech-to-Text
Convert speech content into text, suitable for scenarios such as meeting minutes and subtitle generation.
Word Error Rate as low as 2.99%
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase