# Acoustic analyses of the RAVDESS corpus of emotional stimuli

## TL;DR

This paper analyzes the acoustic properties of speech in an emotional speech database to help researchers better interpret emotion perception studies.

## Contribution

The study provides a detailed analysis of prosodic cues in the RAVDESS database, which is often missing in emotional stimuli research.

## Key findings

- The RAVDESS database contains measurable prosodic differences in fundamental frequency, duration, and loudness across emotions.
- These acoustic cues vary systematically with different emotional expressions in the speech stimuli.
- The findings offer a reference for interpreting emotional perception studies using this database.

## Abstract

Freely available online emotional stimuli allow researchers to conduct emotional perception research without needing to record their own and easily compare findings across studies; however, the acoustic properties, specifically the prosodic cues, are frequently unreported. Prosodic cues are important for a listener to contrast between the talker's emotional tone. Thus, understanding how these cues differ among an emotional stimuli database allows for a nuanced interpretation of findings for emotion perception researchers. This paper analyzes the prosodic cues (fundamental frequency, duration, and loudness) of the speech stimuli in The Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS).