[Not using YMM4] Introducing how to create Zundamon videos (Yukkuri videos) on Mac [Overview]
(Updated February 16, 2026)
The practical guide has been released!
Hello, I am isuko, the manager of the Delicate Gadget Review.
I have been running a blog, and I have also launched a YouTube channel.
I am an introverted HSP, so I do not show my face, and I am not very keen on using my own voice.
Therefore, I intended to create YouTube videos using synthetic voice to provide commentary on my behalf, so-called Zundamon videos.
The mainstream way to create Zundamon or Yukkuri videos is to use Windows software called Yukkuri MovieMaker4 (YMM4).
However, I am a Mac user.
There are several ways to run Windows software virtually on a Mac, but the processing speed inevitably drops, and in that case, it would be better to just use Windows.
So, by researching and testing ways to create Zundamon videos on a Mac, I have become able to create the kind of videos I do now.
In this article, I will introduce an overview of how to create Zundamon videos like the ones I make on a Mac.
How to create Zundamon videos on a Mac
Write a blog post
First, I write a blog post.
Of course, it is also possible to create a script for a YouTube video directly.
Create a YouTube video script based on the blog post
I create a YouTube video script based on the blog post I wrote.
In my case, the text of the blog post becomes the subtitles and narration for the YouTube video almost as it is.
However, since I believe that YouTube video subtitles are easier to read overall if they are kept to one line, I break up long sentences with commas (、) to shorten the content displayed at one time.
Also, I omit periods (。) because I think they are unnecessary.
I used to do this manually, but it was very tedious, so I now utilize generative AI.
Create voiceover audio for YouTube videos
Load the created text into text-to-speech software to create voiceover audio for YouTube videos.
I currently use the paid app VOICEPEAK.
There is also a similar free app called VOICEVOX.
Since I used VOICEVOX for my early videos, I have experience with both.
As a result, I feel that VOICEPEAK has more natural intonation by default, which allows me to save time on work.
Add silent segments to the exported audio
Personally, I feel that the intervals in the raw exported audio are too short, so I add silent segments.
With both VOICEPEAK and VOICEVOX, audio can be exported in individual blocks that are easy to read.
I add silent segments to each of these audio files.
However, I vary the length of the silent segments between parts of a sentence separated by commas (、) and between sentences separated by periods (。).
For this task, I use a Python program created with the help of generative AI.
Create an fcpxml file based on audio and text
With both VOICEPEAK and VOICEVOX, you can export audio and text block by block.
Based on these files, I create an fcpxml file that can be imported into Final Cut Pro.
By importing the fcpxml created this time into Final Cut Pro, I can start editing with the subtitles and audio already placed on the timeline, which is extremely efficient.
If you are not creating subtitles, you can simply place the created audio files into your preferred video editing app.
However, since synthesized speech often has unnatural intonation, I believe subtitles are necessary.
But placing subtitles manually is extremely tedious.
Therefore, I use a Python program created with the help of generative AI to generate an fcpxml file that places the corresponding text as subtitles according to the length of the audio.
Open the fcpxml file in Final Cut Pro and add insert footage, etc.
Open the created fcpxml file in the video editing software Final Cut Pro and add insert footage, ending footage, etc.
In my case, I also insert audio recorded to verify earphone microphone quality, or videos of me typing on the keyboard.
Record video that matches the playback of the read-aloud audio
In my case, I create the read-aloud audio first and then record the video to match that audio.
If you already have the video, please be mindful of the length of the read-aloud audio.
Also, since I sometimes notice content I forgot to mention or mistakes at this stage and correct them, I create the character animation video after this.
Export only the audio of the video
To create the character animation video, I export the audio of the video.
Since insert footage, ending footage, etc., have been added to the audio exported from VOICEPEAK or VOICEVOX, I output the audio again.
Create character animation video from the exported audio
For the character animation video, in my case, I make the character lip-sync when there is sound and keep the mouth closed when it is silent.
I also make it blink somewhat randomly.
I create this video using a Python program developed with the help of generative AI.
I prepare four patterns of character images for both the mouth-open/closed state and the eye-open/closed state, and by having the program load the audio data exported earlier, I create a character video that matches the length of the video.
Edit and export the video
Arrange the recorded video, the created character video, BGM, sound effects, etc.
I think you need to remove the background from the character video.
Finally, export the edited video.
Summary
In this article, I provided an overview of how to create Zundamon videos like the ones I make on a Mac.
This is just an overview, and I think creating the fcpxml file in particular is difficult.
The Python program I currently use has been customized for my own needs, so it is difficult to share it as is.
However, if there is a lot of demand from people saying, 'I don't mind paying, I want to use that program, and I want to know the specific setup steps,' I am considering cleaning up the code so that anyone can use it and releasing it.
For those who want to efficiently create Zundamon videos on a Mac, please show your support by clicking the 'Like' button or sharing this with others.
If there are many 'Likes,' I will begin writing the sequel (the practical guide).
I also look forward to your comments, questions, and feedback.