Synthesis of fast speech with interpolation of adapted HSMMs and its evaluation by blind and sighted listeners