Text this: Channel compensation of speaker identification based on maximum a posteriori