from optimum.intel.openvino import OVModelForAudioClassification
from transformers import AutoFeatureExtractor, pipeline

# model_id should be set to either a local directory or a model available on the HuggingFace hub.
model_id = "helenai/MIT-ast-finetuned-speech-commands-v2-ov"
feature_extractor = AutoFeatureExtractor.from_pretrained(model_id)
model = OVModelForAudioClassification.from_pretrained(model_id)
pipe = pipeline("audio-classification", model=model, feature_extractor=feature_extractor)
result = pipe("https://datasets-server.huggingface.co/assets/speech_commands/--/v0.01/test/38/audio/audio.mp3")
print(result)