https://blog.mickeyzzc.tech/en/tags/reproducible/Recent content in Reproducible 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-4/Sat, 11 Jul 2026 00:00:00 +0000https://blog.mickeyzzc.tech/en/posts/aihelper/ai-video-audio-synthesis-4/<h2 id="overview">Overview</h2> <p>This is the final installment of the series. The previous three parts covered generating footage with Remotion, batch voiceover with edge-tts, and offline BGM synthesis with numpy. This part brings everything together: using <strong>ffmpeg filter_complex</strong> to mux the silent video, 7 voice clips, and one BGM into the final export — along with the end-to-end workflow, a pitfall cookbook, and the underlying principles.</p> <p>After reading, you will understand:</p> <ul> <li>Why <code>-c:v copy</code> is much faster and lossless compared to re-encoding</li> <li>How to arrange multiple voice clips on a sequential timeline to avoid overlap</li> <li>The mathematical meaning of each filter in <code>filter_complex</code></li> <li>How to batch-produce 6 language variants with one command</li> </ul> <h2 id="ffmpeg-muxing">ffmpeg Muxing</h2> <h3 id="overall-approach">Overall Approach</h3> <p>The muxing stage has a simple job: pack three things into one MP4.</p>