https://blog.mickeyzzc.tech/en/tags/remotion/Recent content in Remotion on Mi&Bee BlogHugo -- gohugo.ioen蓝宝石的傻话Sat, 11 Jul 2026 00:00:00 +0000https://blog.mickeyzzc.tech/en/posts/aihelper/ai-video-audio-synthesis-1/Sat, 11 Jul 2026 00:00:00 +0000https://blog.mickeyzzc.tech/en/posts/aihelper/ai-video-audio-synthesis-1/<p>This article is based on hands-on experience from the MiBee NVR open-source 45-second promo video project. You will learn how to <strong>generate video footage by code</strong> (Remotion), <strong>produce AI voiceovers</strong> (edge-tts), <strong>synthesize BGM offline</strong> (numpy), and <strong>mux everything into a final video</strong> with ffmpeg. All steps are ready to follow.</p> <h2 id="what-this-technology-does">What This Technology Does</h2> <p>A multilingual promo project consists of three independent production stages, finally muxed by ffmpeg:</p> <table> <thead> <tr> <th>Stage</th> <th>Tool</th> <th>Output</th> </tr> </thead> <tbody> <tr> <td>Footage</td> <td><strong>Remotion</strong> (React-based video)</td> <td>Silent MP4</td> </tr> <tr> <td>Voiceover</td> <td><strong>edge-tts</strong> (Microsoft free TTS)</td> <td>MP3 per clip</td> </tr> <tr> <td>BGM</td> <td><strong>numpy</strong> offline synthesis</td> <td>WAV file</td> </tr> <tr> <td>Muxing</td> <td><strong>ffmpeg</strong> filter_complex</td> <td>Final MP4 (video + voice + BGM)</td> </tr> </tbody> </table> <p>The overall pipeline looks like this:</p>