Home
Welcome to the Portrait Synthesis from Speech wiki!
This is a research project associated with Retratista application for face generation and identification from speech (text).
The software uses a description of a person's face and turns it into a realistic portrait based on the description. This description can be provided by means of speech or textual description or even manual attributes selection using sliders, checkboxes ..., etc. The generated face can be refined iteratively, in order to, precisely, match the description. Also, the software can generate different poses of the generated face to further improve identification.
This is helpful for criminal or missing individual identification.
Figure(1): Complete system architecture diagram showing the flow between different modules.