<![CDATA[Since COVID-19, Zoom has become the primary distance communication platform integrated with automatic transcription tools such as Otter and Descript. However, meetings involving non-native French speakers can generate sound interference that reduces transcription accuracy. This study compares the automatic French transcription quality of both tools in corporate online meetings affected by interference from native Indonesian speakers. The data consists of online meeting recordings of Indonesian speakers in French based on the textbook, Affaires.com. The analysis applies Weinreich’s (2010) interference theory, Le Querler’s (1994) word class theory, Word Error Rate, and Tholly et al.’s (2025) severity levels of automatic French transcription errors. The results show that both tools produced accurate transcriptions in the 10% error range, with Otter (10.18%) being slightly more accurate than Descript (10.22%). Regarding interference, phonetic–non-phonemic errors were the easiest to handle, while phonological–grammatical errors were the most difficult for both tools. Most of Otter’s errors fall into the most severe Fail category, causing incomprehensibility, influenced by its stronger language understanding. Meanwhile, Descript shows more Gram-level errors, which remain understandable but reflect weaker language understanding. Therefore, Otter is recommended for its stronger language comprehension, whereas Descript is recommended because its errors are still intelligible.]]>
Copyrights © 2026
