Vizard is a hosted clipper with a clean dashboard. Podcli is the local-first open-source version: no monthly cap, no watermark, no upload, scriptable from the CLI.
The features that change the day-to-day for clip creators.
Yes. Podcli is MIT licensed and free. There is no per-minute pricing, no plan tiers, and no watermark. The only optional cost is the API key you bring for AI clip suggestion (Claude or OpenAI), and that step is skippable.
Vizard and most hosted clippers price around monthly processing minutes. If your show is 90 minutes a week and you batch a few episodes at once, you hit the cap fast. Podcli has no concept of a minute cap. The work happens on your laptop. The cost is wall-clock time plus any AI API calls you choose to make.
Vizard ships a template library you pick from. Podcli ships four styles (branded, hormozi, karaoke, subtle), each a real React component rendered by Remotion.
For most shows the four defaults cover what you would pick anyway. For shows that want a unique signature look, editing JSX beats picking a preset.
Vizard does not expose an MCP server. Podcli does. An AI coding agent (Claude Code, Codex, Cursor) can run the whole pipeline for you, including titles and descriptions, while you do other work. For people already living inside an AI coding tool, this changes the workflow.
The honest version. Steps in the order you'd actually do them.
Install: git clone https://github.com/nmbrthirteen/podcli && cd podcli && ./setup.sh.
Use the web UI or CLI the same way you would use Vizard's dashboard, but pointing at local files.
Pick a caption style. The four built-ins cover the look-and-feel of Vizard's defaults.
Optional: fill in .podcli/knowledge/ with your show identity. This replaces Vizard's brand kit.
Optional: add an API key for Claude or OpenAI to get AI clip scoring.
Run a batch. Output: 1080x1920 MP4s ready to upload.
Direct answers to the searches people run before they decide.
Yes. Podcli is free under MIT. There is no paid tier. You only pay for an AI API key if you opt in to AI clip suggestions.
Yes. Podcli uses YuNet for face detection and a per-clip mouth-motion analysis to track the active speaker on split-screen interviews. No diarization is required for 2-person split-screen.
Yes. The CLI and MCP server both support batch processing. There is no monthly cap because nothing runs in the cloud.
Whisper supports 99 languages including Spanish, Portuguese, French, German, Mandarin, Hindi, and more. Captions and transcripts work in any of them.
No. Hardware encoders are used when available (VideoToolbox on Mac, NVENC on NVIDIA GPUs, VAAPI on Linux), but the CPU fallback works on any modern laptop. Whisper transcription is the only step that benefits from a GPU and even there CPU works fine.
Open source, MIT, no signup, no watermark, no upload. Clone and run.