YPP
Unlike automating a blog without Google Ads
My YouTube channel has enabled YPP (YouTube Partner Program), so theoretically, as long as there’s a certain amount of viewership, there will be some income.
Unfortunately, I don’t have a graphics card. Recently, Twitter keeps mentioning an open-source AI project for fully automated short video creation; otherwise, I could give it a try.
So currently, I can only test long video automation, not content generation. Instead, I’m using a workflow to automate tasks like subtitles, voiceovers, and background noise removal.
Automation
My original long video production involved editing with CapCut.
Using it to remove vocals while retaining background audio.
Using it to generate bilingual subtitles.
Using it to generate AI voiceovers based on subtitles.
Using it to merge and export videos.
By the way, all the above AI capabilities are VIP-only. Editing these usually takes me over half an an hour, and the results aren’t ideal.
Now, I’m deploying Fast Whisper for speech-to-text (transcription).
Using other open-source tools to generate TTS (Text-to-Speech) audio from subtitles. Currently, I’m not using voice cloning; there’s no need.
Then, I remove my original voice while keeping the background sound. There’s an open-source project for this too, but I don’t remember its name.
Finally, I use FFmpeg to merge everything into a single video and save it to a folder.
My videos are self-shot. For content, I use a local large language model to read blog posts (like this one), organize them into subtitles, automatically generate voiceovers, and automatically trim the video length to match the audio.
This is like adding a podcast or a roughly processed video to writing articles.
The entire workflow takes 10 minutes.
I plan to upload the generated videos to YouTube and then embed them into blog post video addresses, like for this article.
Then, I’ll test YouTube video earnings.
Video Content
Of course, this is currently just an experiment, a test.
Later, I will consider optimizing and adjusting the videos to be PPTs, screen-recorded courses, and some interesting or practical content.
For now, I just want to do faceless voiceovers, expanding text content in video form.
This current workflow:
- Eliminates VIP fees for editing software.
- Eliminates half an hour or more of manual editing.
- Expands blog posts into videos, achieving mutual supplementation between articles and videos.
Earnings Statistics
The blog website is still in the indexing and traffic generation phase.
YouTube directly tracks earnings.
Similarly, if this method is successful, other channels can also refer to the same approach for publishing long videos.
It can be described as semi-automatic, with original videos, just integrating some AI into the automation.
Currently, I set the video length to be over 3 minutes. Of course, the content still relies on the blog post content.
If the blog post content is scarce, even AI can’t create something out of nothing.
Currently, I plan to blog daily about my attempts at automation and automation-related money-making.
One is for my own convenience in organizing, recording, and optimizing.
The other is for sharing, making it easier for everyone to discuss, exchange ideas, or refer to.
This is not a tutorial after success, but a daily log of actual tests, often featuring scenarios of failure.
As well as some thoughts and ideas; thinking is very important.
Currently, my main income comes from YPP. Although it’s not enough to buy a cup of coffee a month, it doesn’t hinder testing.
Traffic and algorithms are inherently a matter of luck.