Noise in Speech-to-Text Voice: Analysis of Errors and Feasibility of Phonetic Similarity for Their Correction