Focus Areas: AI Model Evaluation, Bengali Language Quality, Audio & Video Annotation, Translation Accuracy, TTS, Dubbing & Quality Assurance

I work on evaluating AI-generated outputs across text, audio, and video, applying task-specific guidelines to assess quality, accuracy, naturalness, and alignment with the original input.

Core Responsibilities:

• Evaluate AI-generated text, audio, and video outputs against defined quality criteria
• Assess translation accuracy, meaning preservation, and language naturalness
• Compare text-to-speech (TTS) and dubbing outputs for voice similarity, pronunciation, pacing, and prosody
• Review audio-video alignment, lip-sync, and speech consistency where required
• Correct transcripts and verify segment text, timestamps, and alignment with recorded speech
• Perform structured side-by-side evaluations and provide evidence-based ratings
• Follow project guidelines, maintain annotation consistency, and identify quality issues

Evaluation Dimensions:

• Translation Accuracy & Meaning Preservation
• Voice Similarity & Speaker Consistency
• Speech Naturalness & Native Fluency
• Pronunciation, Pacing & Prosody
• Audio Quality & Clarity
• Transcript Accuracy & Timestamp Alignment
• Lip-Sync & Audio-Visual Consistency
• Overall Output Quality & Preference

Selected Evaluation Work: