Linguistic Data Annotator for AI training speech annotation (conversational English)
Annotated conversational English speech data for AI training with strict adherence to annotation guidelines. Produced time-aligned verbatim orthographic transcriptions capturing fillers, false starts, and overlapping speech. Validated and corrected segmentation boundaries and prosodic annotations to improve linguistic consistency across large datasets. • Reviewed automated speech segmentation for intonational phrase boundary accuracy. • Identified prosodic features including pitch resets, stress patterns, and final syllable lengthening. • Checked waveform/spectrogram alignment in Praat against spoken utterances. • Ensured consistent transcription conventions across datasets.