finds.dev← search

// the find

harry0703/MoneyPrinterTurbo

★ 99,130 · Python · MIT · updated Jul 2026

利用 AI 大模型和自动化工作流,根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.

MoneyPrinterTurbo automates short-form video production end-to-end: give it a topic, it generates a script via an LLM, pulls stock footage from Pexels/Pixabay/Coverr, synthesizes voiceover, burns in subtitles, and optionally posts to TikTok/Instagram/YouTube. It's aimed at people running content channels at volume, not at filmmakers or anyone who cares about craft.

The provider abstraction is genuinely well done — you can swap LLMs (OpenAI, Gemini, DeepSeek, local Ollama) and TTS engines (Edge TTS free tier, ElevenLabs, Azure) without touching code, just config. The two-mode subtitle system is a smart tradeoff: fast TTS-timestamp alignment for most cases, Whisper transcription when you need frame-accurate timing. Deployment story is better than average for a Python project of this age: Docker image on GHCR, uv support, a Windows one-click package, and Colab for zero-setup tryouts. The test suite is surprisingly thorough — there are actual unit and integration tests for most services, which is rare in projects like this.

Footage matching is pure keyword extraction: the LLM picks search terms, the stock API returns clips, and you get whatever comes back — there's no semantic relevance check between the clip and the sentence it's supposed to illustrate, so the visual non sequiturs are frequent and obvious. The README buries technical content under a wall of sponsor blocks (seven sponsors with referral links before you reach the feature list), which tells you something about the project's priorities. The social posting integration depends entirely on a third-party paid service (Upload-Post) that the project has no control over, so that whole feature can break at any time for non-technical reasons. And despite the 'one-click' framing, getting all the moving parts working — LLM key, TTS key, stock footage key, ffmpeg, optionally Whisper model — is a multi-hour setup for anyone who hasn't done it before.

View on GitHub →

// want more like this?

We dig through GitHub every week and send a few repos picked for what you actually care about — each with an honest take like this one.

Get finds in your inbox → Search again →