The paper discusses the way in which speech sounds are being evaluated and transcribed in various fields of speech sciences, and suggests ways for a more accurate transcription. The academic fields explored are of phonetics, speech processing, speech pathology, and foreign language education. The discussion centers on the International Phonetic Alphabet (IPA), most commonly used in these fields, and other less widely-accepted transcription conventions such as the TOnes and Break Indices (ToBI), the Speech Assessment Methods Phonetic Alphabet (SAMPA), an extension of the official Korean Romanization (KORBET), and the American-English transcription system in the TIMIT database (TIMITBET). These transcription conventions are dealt with Korean, English, and Korean-accented English. The paper demonstrates that each transcription can exclusively be recommended for a specific need from different academic fields. Due to its publicity, the IPA is best suited for phonetic evaluation in the fields of phonetics, speech pathology, and foreign language education. The rest of the transcriptions are useful for keyboard-inputting the phonetically evaluated data from all these fields as well as for sound transcription in speech engineering, because they use convenient letter symbols for typing, searching, and programming. Several practical suggestions are made to maintain the transcriptional efficiency and consistency to accommodate the intra-and inter-transcriber variability.