Acoustic models of Brazilian Portuguese Speech based on Neural Transformers