Application of Automatic Speech Recognition to Quantitative Assessment of Tracheoesophageal Speech with Different Signal Quality