Podcast Light Edit Chain
One-command light edit for spoken-word recordings: cut ums with the Uhm frame classifier (no ASR), shorten long pauses, voice-fitted EQ, -16 LUFS. With a measured demo.
About This Project
Podcast-Light-Edit-Chain provides a one-command workflow to prepare solo spoken-word recordings for publication. By leveraging the Uhm frame classifier to identify fillers without the overhead of automatic speech recognition, the tool removes disfluencies and applies crossfades. It further refines audio quality through adaptive silence gating, voice-specific equalization based on long-term spectral analysis, and standardized loudness normalization to -16 LUFS.
The project is designed for efficiency and transparency, generating detailed JSON logs for every edit to ensure that automated changes can be audited or replayed. Built with Python and utilizing FFmpeg for processing, it is intended for creators looking to streamline the post-production of podcasts, voice notes, or monologues while maintaining high precision and control over the final output.