# Emotion Dependent Facial Animation from Affective Speech

## TL;DR

This paper introduces a two-stage deep learning method for emotion-dependent facial animation driven by affective speech, improving realism by classifying emotions and then synthesizing facial shapes accordingly.

## Contribution

It presents a novel emotion-dependent facial animation framework that outperforms universal models by classifying emotions before synthesis.

## Key findings

- Emotion classification accuracy improves facial animation quality.
- Emotion-specific models outperform universal models in MSE loss.
- Subjective evaluations favor emotion-dependent over generic models.

## Abstract

In human-to-computer interaction, facial animation in synchrony with affective speech can deliver more naturalistic conversational agents. In this paper, we present a two-stage deep learning approach for affective speech driven facial shape animation. In the first stage, we classify affective speech into seven emotion categories. In the second stage, we train separate deep estimators within each emotion category to synthesize facial shape from the affective speech. Objective and subjective evaluations are performed over the SAVEE dataset. The proposed emotion dependent facial shape model performs better in terms of the Mean Squared Error (MSE) loss and in generating the landmark animations, as compared to training a universal model regardless of the emotion.