Speech stress-detection

The goal of the stress-detection project is to identify various emotions in speech audio samples.

While much can be infered from the textual content and context of speech, this project focusses on emotion identification at the spoken utterance level, examining audio signals independently from transcription.

Being able to distinguish between stressed and neutral versions of the same phrase is imperative in communication applications where a "calm" tone of voice is encouraged. Thus, our approach is to look past the words for physiological signs of emotion within speech.