Deterministic Multimodal Quality Control for Generative Video Pipelines: Automated Word-Level Whisper Alignment, Sub-Frame Boundary De-Glitching, and Non-Destructive NLE Synthesis
Technical specification and empirical validation of a deterministic multimodal Quality Control (QC) pipeline for generative AI video production (Google Flow / Veo) by HM Shihabuddin Jashory (Shihab). Details automated word-level Whisper timestamp alignment, cut-boundary audio clipping detection, 1-second contact sheet visual auditing, 0.5-second close-up lip-sync isolation verification, and non-destructive CapCut PC NLE draft synthesis.
Authors
- HM Shihabuddin Jashory (ORCID: https://orcid.org/0009-0003-4163-0092)
Institutions
- University of the People (US)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-31
- DOI
- https://doi.org/10.5281/zenodo.23832730
- Primary Topic
- Multimedia Communication and Technology
- Type
- article
- Field-Weighted Citation Impact
- 0.00