Skip to content

Latest commit

 

History

History
2 lines (2 loc) · 351 Bytes

File metadata and controls

2 lines (2 loc) · 351 Bytes

Art-generation-using-speech-emotions

We have proposed an image generation approach using speech emotion recognition. We have used VQGAN and CLIP models in tandem to generate an image from a text prompt from speech and emotions recognized from the speech spectrogram. We see that the achieved accuracy of the emotion recognition model was about 75%.