Swahili Annotation
Audio
TranscriptionThe project focused on Swahili data annotation. My role involved listening to audio recordings and refining AI-generated transcriptions to ensure they accurately reflected what was spoken. I was also responsible for segmenting the audio and labeling speakers to make them easily identifiable. The project ran for three months and included a variety of audio files ranging from 10 minutes to 1 hour in length, each with different lengths. For quality assurance, the work underwent a two-stage review process. After the first inspection, any identified errors were returned for correction before resubmission. The corrected tasks were then reviewed again twice for final approval and successful completion.
2025 - 2025