AI Data Annotator - Speech and Audio (Appen Jigglypuff Project)
Reviewed and accurately annotated machine-generated transcripts of conversational Swahili audio, applying taxonomy rules to label filler words, stutters, and reduced speech forms. Categorized non-speech sounds using a predefined tag library and applied structured overlap annotation guidelines for simultaneous speech. Consistently flagged ambiguities, maintained strict accuracy targets, and ensured rubric compliance throughout all tasks. • Placed and resized audio segment chips on speaker-specific waveform lanes and applied locale-based speaker tagging. • Verified disfluency detections using automated tools, correcting errors for optimal training dataset quality. • Committed 10–20 hours per week while exceeding participation and accuracy standards for the Appen Jigglypuff Project. • Utilized ADAP by Appen and waveform editing tools to ensure high-quality, structured annotation outputs.