Why Inference Time of Whisper-Small-Int8-Dynamic-Inc Model Was Faster Than Whisper-Base Model?

• Exported and converted whisper-small-int8-dynamic-inc into INT8 ONNX model
• Exported and converted whisper-base into INT8 OpenVINO™ model
• Inferred whisper-small-int8-dynamic-inc and whisper-base model
• Inference time of whisper-small-int8-dynamic-inc model was faster than whisper-base model

whisper-small-int8-dynamic-inc model was based on whisper-small model which is different from whisper-base model. Therefore, the inference time will be different when comparing whisper-small-int8-dynamic-inc model and whisper-base model.