How to build SER model?

I want to build a speech emotion recognition model that correctly predicts the emotion by speech. Is there any pretrained models like speechbrain,Wav2Vec2. If any provide the correct model name like name of Wav2Vec2 model or is it better to build the model from scratch like building a CNN model from scratch or finetuning ASR pretrained model is better.Give me suggestions I need minimum 80% accuracy

Apr 15, 2025 - 06:35

I want to build a speech emotion recognition model that correctly predicts the emotion by speech. Is there any pretrained models like speechbrain,Wav2Vec2. If any provide the correct model name like name of Wav2Vec2 model or is it better to build the model from scratch like building a CNN model from scratch or finetuning ASR pretrained model is better.Give me suggestions I need minimum 80% accuracy