metadata

language:
  - bn
metrics:
  - accuracy
pipeline_tag: audio-classification

whisper-tiny-bn-emo

This model is a fine-tuned version of shhossain/whisper-tiny-bn on the Unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.1842 Accuracy: 0.9357

Model Info

It detects 7 basic human emotions on Bengali Language.

from transformers import pipeline

pipe = pipeline("audio-classification", model="shhossain/whisper-tiny-bn-emo")

pipe("audio_file.wav")