Audio Annotation Specialist (AI Data Solutions, Contract)
Transcribed and labeled 1,500+ hours of speech audio to support Automatic Speech Recognition (ASR) model training while maintaining accuracy rates above 97%. Performed speaker diarization and tagged non-speech audio events such as background noise, music, and environmental sounds using a structured taxonomy. Completed quality assurance by reviewing and correcting peer annotations to keep labeling consistent with project guidelines. • Speech transcription for ASR dataset creation • Speaker diarization for multi-speaker recordings across accents/noise • Sound event tagging for non-speech environmental audio • Peer annotation review and QA feedback loops