We are seeking professional-level French speakers with intermediate to expert proficiency in ARPABET phonemic transcription. Candidates must be able to accurately identify reading errors, pronunciation variations, and speech disfluencies. Experience in phonetics, linguistics, automatic speech recognition (ASR) datasets, or child-speech transcription is strongly preferred. Applicants must be available to start within one week, pass a qualification assessment, and commit at least 20 hours per week.
In this role, you will listen closely to French audio clips (typically 1–2 minutes each) and validate existing word-level and phoneme-level transcriptions against the audio using a web-based annotation platform. Your primary responsibility will be to verify transcription accuracy, identify and correct discrepancies, and ensure that reading errors, pronunciation variations, and speech disfluencies are accurately represented. You will not be required to transcribe audio from scratch. Work is assigned in ongoing batches, with annotations reviewed through a peer quality assurance process before final approval. Your contributions will support the development of advanced speech recognition and language-learning AI models. Payment is based on approved tasks, with greater accuracy and productivity offering higher earning potential.