What's the best way to 'normalize' a string in Redshift?

Question

Since my texts are in Portuguese, there are many words with accent and other special characters, like: "coração", "hambúrguer", "São Paulo".