metadata
license: mit
language:
- ru
base_model:
- facebook/wav2vec2-base
pipeline_tag: audio-classification
tags:
- speech
datasets:
- xbgoose/dusha
library_name: transformers
Model Description
This model is used for BMSTU course work: Web-application for emotion recognition in speech.
Direct Use
This model can recognize emotions (perfectly on russian language speech) without extracting context of speech. Classes of emotions:
- positive;
- sad;
- angry;
- neutral;
- other (mostly used for noises)