At RWS, I audit annotator evaluations of text-to-speech outputs, checking pronunciation, audio quality, naturalness, and speaker similarity against detailed rubrics. I’ve completed 500+ TTS audits across 50 hours and document judgments with clips, timestamps, and reasons.
At RWS, I also audit evaluations of live multilingual speech-to-speech conversations. I’ve completed 400+ conversation audits across 100+ hours, checking ratings and missed or falsely flagged issues such as latency, hallucination, and interruption.
As an AI Response Evaluator at Outlier, I evaluated voice AI outputs and executed red-team scenarios. I also transcribed Bengali audio and video at GoTranscript, and use Praat for acoustic and phonetic analysis.

