Audio and Speech Labelling Project
Worked on an audio data labelling project involving transcription of spoken recordings across multiple accents and speaking styles. Tasks included tagging speech segments with metadata such as speaker identity, emotion (happy, sad, angry, neutral), tone, and background noise levels. Reviewed and quality-checked 500+ audio clips to ensure accuracy met client specifications, achieving a 98% approval rate. Followed strict annotation guidelines to maintain consistency across all delivered datasets.