French Phonetic Transcription for Speech Recognition Model
As a freelance contributor for Outlier AI, I participated in a project focused on enhancing the accuracy of a French speech recognition system. The main task was to transcribe audio clips from native French speakers and convert them into precise phonetic representations using the ARPABET transcription system. My responsibilities included: Listening to short audio samples (1–10 seconds) and accurately transcribing the spoken content. Applying ARPABET phonetic labels to each word or syllable with precision. Annotating speaker turns, background noise, and any mispronunciations when detected. Reviewing and correcting phonetic inconsistencies to ensure adherence to quality standards. The objective of the project was to train a multilingual AI model to better recognize variations in French pronunciation. During this assignment, I completed over 1,000 labeled audio segments, maintaining an accuracy rate above 95%.