<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Tinkering Notes on yexca'Blog</title><link>https://blog.yexca.net/en/categories/tinkering-notes/</link><description>Recent content in Tinkering Notes on yexca'Blog</description><generator>Hugo -- gohugo.io</generator><language>en</language><copyright>yexca</copyright><lastBuildDate>Mon, 22 Jun 2026 01:26:59 +0900</lastBuildDate><atom:link href="https://blog.yexca.net/en/categories/tinkering-notes/feed.xml" rel="self" type="application/rss+xml"/><item><title>Remote Control Your Windows PC</title><link>https://blog.yexca.net/en/archives/287/</link><pubDate>Mon, 22 Jun 2026 01:26:59 +0900</pubDate><guid>https://blog.yexca.net/en/archives/287/</guid><description>📢 This article was translated by gemini-3.5-flash Lately, when my local PC is training models, it completely locks me out of doing other work. Using my Mac or Linux devices feels a bit off too, since they already have their own dedicated tasks. So, I figured I’d put the lab computer to use. Once set up, I can run my training there and keep doing my own thing on my main machine.</description></item><item><title>Deploying Local Uncensored AI Models: Breaking the Two Layers of Shackles</title><link>https://blog.yexca.net/en/archives/286/</link><pubDate>Tue, 16 Jun 2026 13:07:08 +0900</pubDate><guid>https://blog.yexca.net/en/archives/286/</guid><description>📢 This article was translated by gemini-3.5-flash I used to think locally deployed models could speak completely freely. But after hands-on experience, I realized modern AI models have two layers of shackles: one is the cloud provider’s external filter, and the other is the “refusal neurons” deeply baked into the model weights during safety alignment training. To get the perfect experience, you need to deploy an uncensored model.
Warning Uncensored versions generally remove …</description></item><item><title>Local Deployment of ZeroClaw</title><link>https://blog.yexca.net/en/archives/285/</link><pubDate>Sat, 13 Jun 2026 13:02:57 +0900</pubDate><guid>https://blog.yexca.net/en/archives/285/</guid><description>📢 This article was translated by gemini-3.5-flash Back when OpenClaw was super popular, I wasn’t really a fan of using Node for deployment. The subsequent security vulnerabilities confirmed my doubts. Now, there are plenty of great forks. Since I prefer lightweight setups, I went with the lighter ZeroClaw.
Since I’m deploying my own AI assistant, I don’t want my data uploaded to the cloud. So, the goal this time is to build a fully local private assistant.</description></item><item><title>First Try at Vibe Coding: Accidentally Inhaling a New Digital Drug</title><link>https://blog.yexca.net/en/archives/284/</link><pubDate>Sun, 07 Jun 2026 18:42:10 +0900</pubDate><guid>https://blog.yexca.net/en/archives/284/</guid><description>📢 This article was translated by gemini-3.5-flash A friend strongly recommended Codex to me recently, so I gave it a spin. After trying it myself, I’d call it an “idea extractor.”
Before this, my impression of AI coding was that it could build things, but the resulting code would be a mess to maintain. But after actually getting my hands dirty, I burned through 72M tokens in one go. My ideas kept getting pulled out and turned into reality. It was a massive productivity …</description></item><item><title>Discussion on Audio Preprocessing for AI Cover Training</title><link>https://blog.yexca.net/en/archives/283/</link><pubDate>Tue, 02 Jun 2026 17:23:44 +0900</pubDate><guid>https://blog.yexca.net/en/archives/283/</guid><description>📢 This article was translated by gemini-3.5-flash Completely relying on AI to extract clean vocals is still not quite realistic with current technology; in most cases, manual correction is needed. However, as long as you have enough data, you can always scrape together enough AI-extracted vocals suitable for training.
This article discusses a workflow for filtering training audio using only AI, from the perspective of an outsider. Everything used here is open-source.</description></item><item><title>Intranet Penetration with frp</title><link>https://blog.yexca.net/en/archives/281/</link><pubDate>Thu, 28 May 2026 15:52:18 +0900</pubDate><guid>https://blog.yexca.net/en/archives/281/</guid><description>📢 This article was translated by gemini-3.5-flash Set up intranet penetration using frp with a full Docker deployment. Tested the speed, and it’s pretty good with acceptable latency.
DNS Configuration You’ll need a server with a public IP. I handle my SSL certificates via Cloudflare, so all my connections go through port 80.
Configure a few domains pointing to your server: one for connecting to frp (e.g., frp.example.com), and others for the services you want to proxy (e.g., …</description></item><item><title>AI Covers with Roformer and Applio</title><link>https://blog.yexca.net/en/archives/280/</link><pubDate>Sat, 23 May 2026 03:00:06 +0900</pubDate><guid>https://blog.yexca.net/en/archives/280/</guid><description>📢 This article was translated by gemini-3.5-flash Getting Audio For training, real lossless audio is highly recommended (specifically pcm_f32le WAV files). It’s best to rip them from physical albums. You can refer to this guide: https://www.bilibili.com/opus/925630344961458181 , but skip the FLAC conversion and export directly to WAV.
This guide uses Applio (based on the RVC architecture). You’ll need about 10 to 60 minutes of clean vocals (acapella) to get good training …</description></item><item><title>Common FFmpeg Commands</title><link>https://blog.yexca.net/en/archives/276/</link><pubDate>Wed, 15 Apr 2026 09:27:01 +0900</pubDate><guid>https://blog.yexca.net/en/archives/276/</guid><description>📢 This article was translated by gemini-3-flash-preview When it comes to audio and video processing, FFmpeg is the ultimate open-source tool. I’ve been using it for a while, but my notes were scattered. Since I’m using it more frequently now, I decided to compile my most-used commands here (and add a few missing ones).
Convert Video Format 1 ffmpeg -i video.mp4 video.avi Extract Audio from Video 1 ffmpeg -i input.mp4 -vn output.mp3 -vn: Disables video recording in the output …</description></item><item><title>Your next NAS is not a NAS</title><link>https://blog.yexca.net/en/archives/261/</link><pubDate>Thu, 13 Nov 2025 17:57:29 +0900</pubDate><guid>https://blog.yexca.net/en/archives/261/</guid><description>📢 This article was translated by Gemini-2.5-pro Introduction Back in 2022, when I was first messing with Linux(Simplify Chinese) , I had this idea that you don’t necessarily have to pay for a special OS just to build a NAS.
Fast forward to today. While tinkering with my server, the mount speeds on my cloud drives have been just… disappointing. On impulse, I ordered an N5095 motherboard. I realized the 4GB RAM stick I have lying around is a bit low, but I really didn’t want to …</description></item><item><title>Tinkering with Manga and Audio Websites</title><link>https://blog.yexca.net/en/archives/258/</link><pubDate>Sun, 05 Oct 2025 01:58:35 +0900</pubDate><guid>https://blog.yexca.net/en/archives/258/</guid><description>📢 This article was translated by gemini-2.5-flash Introduction Google’s Gemini Pro subscription comes with 2TiB of space, which naturally got me thinking about tinkering again. So, let’s get to it.
But looking back, it’s already been three years since my last tinkering . Time really flies, huh?
Manga: Komga First up is Komga for reading manga. While scanning can be a bit sluggish, the overall user experience is decent.
After all these years, Komga has seen many updates. My …</description></item><item><title>2025 New Server Deployment Log</title><link>https://blog.yexca.net/en/archives/257/</link><pubDate>Fri, 03 Oct 2025 10:22:25 +0900</pubDate><guid>https://blog.yexca.net/en/archives/257/</guid><description>📢 This article was translated by gemini-2.5-flash Intro My server’s almost up. Renewed it last year, but this year, with all the deals popping up, renewing felt a bit pricey. After checking out some smaller server providers, I saw Alibaba Cloud offers discounts if you’ve been bill-free for 12 months. Since I was already looking at big names, Oracle popped into my head.
Figured I’d give it a shot. Used all my real info, and it actually got approved directly. Sweet!</description></item><item><title>Get OneDrive Direct Download Links</title><link>https://blog.yexca.net/en/archives/237/</link><pubDate>Sun, 02 Mar 2025 12:58:57 +0900</pubDate><guid>https://blog.yexca.net/en/archives/237/</guid><description>📢 This article was translated by gemini-3-flash-preview Introduction Recently, I found that IDM couldn’t automatically capture files shared via OneDrive. Browser downloads were unstable and kept failing, so I decided to find a way to get direct download links.
Extension Issue I initially thought the IDM extension was no longer supported and needed a reinstall. But after deleting it, the Chrome Web Store showed it couldn’t be installed. Well, I deleted it too early.</description></item><item><title>Intranet Tunneling</title><link>https://blog.yexca.net/en/archives/199/</link><pubDate>Fri, 22 Nov 2024 22:05:34 +0800</pubDate><guid>https://blog.yexca.net/en/archives/199/</guid><description>📢 This article was translated by gemini-2.5-flash An old post I wrote, system creation date 2023.12.01.
Looking back at these old posts, I’m a bit hesitant to publish them. But I can’t bring myself to delete them either. After some thought, I’ll just publish them. They’re my own work after all.
ngrok OneClick Go to the official website , register an account, and verify your email to get your Authtoken.</description></item><item><title>2024 PC Setup Log</title><link>https://blog.yexca.net/en/archives/182/</link><pubDate>Wed, 02 Oct 2024 21:35:24 +0800</pubDate><guid>https://blog.yexca.net/en/archives/182/</guid><description>📢 This article was translated by gemini-3-flash-preview Introduction I recently reinstalled my OS after swapping out some drives. When I pulled out my ancient setup USB drive, the software inside was pretty outdated—some wouldn’t even run on a modern system, and for others, I’ve found better alternatives. I’m writing this article to document the software I used for this 2024 build.</description></item><item><title>Cannot Login to Typecho Admin via HTTPS</title><link>https://blog.yexca.net/en/archives/175/</link><pubDate>Fri, 26 Jul 2024 22:22:22 +0800</pubDate><guid>https://blog.yexca.net/en/archives/175/</guid><description>📢 This article was translated by gemini-3-flash-preview Introduction After migrating Typecho to a Docker container, enabling HTTPS caused errors when trying to log into the admin panel. It worked fine over HTTP. Since my previous non-Docker deployment didn’t have this issue, I initially suspected a Docker networking problem. At the time, I only needed to make a one-off update, so I just disabled HTTPS, finished the changes, and left it. Now that I’m using Typecho again and …</description></item><item><title>Installing Xfce on CentOS 7</title><link>https://blog.yexca.net/en/archives/153/</link><pubDate>Sat, 17 Feb 2024 22:51:34 +0800</pubDate><guid>https://blog.yexca.net/en/archives/153/</guid><description>📢 This article was translated by gemini-3-flash-preview When did I even write this? (
Check for Xfce Group 1 yum grouplist If it’s not listed, you need to install the EPEL repository.
1 yum install epel-release -y Install X Window System 1 yum groupinstall &amp;#34;X Window system&amp;#34; Install Xfce 1 yum groupinstall xfce Install Chinese Fonts (KaiTi) 1 yum install cjkuni-ukai-fonts Launch Xfce Desktop 1 systemctl isolate graphical.target References CentOS 7安装Xfce桌面环境过程_qq_28641401的博 …</description></item><item><title>Running Scripts on Fedora Boot</title><link>https://blog.yexca.net/en/archives/149/</link><pubDate>Sun, 28 Jan 2024 14:25:58 +0800</pubDate><guid>https://blog.yexca.net/en/archives/149/</guid><description>📢 This article was translated by gemini-2.5-flash Introduction Seriously, don’t put infinite loops in here!!! We’re using the systemd way.
Create the Script Write your script in this file:
1 vi /etc/rc.d/rc.local Grant execute permissions:
1 chmod +x /etc/rc.d/rc.local Edit Startup Status Edit the file:
1 vi /usr/lib/systemd/system/rc-local.service Add [Install] to the last line:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 # This file is part of systemd. # # …</description></item><item><title>2023 New Server Deployment Log</title><link>https://blog.yexca.net/en/archives/131/</link><pubDate>Fri, 03 Nov 2023 15:34:01 +0800</pubDate><guid>https://blog.yexca.net/en/archives/131/</guid><description>📢 This article was translated by gemini-3-flash-preview After moving my blog to GitHub, I didn’t plan on buying any more servers. Maybe it’s because things have been too quiet lately and I felt like tinkering, or maybe it’s that “I might not need it, but I must have it” mindset, but I ended up making an impulse purchase.
Current Websites (yexca.net domain) Personal Intro (WIP): http://yexca.net This Blog: http://blog.yexca.net VRChat Blog: http://vrc.yexca.net VRChat Docs: …</description></item><item><title>Redirecting a Domain to a URL with a Path Using CNAME</title><link>https://blog.yexca.net/en/archives/116/</link><pubDate>Mon, 28 Aug 2023 15:31:01 +0800</pubDate><guid>https://blog.yexca.net/en/archives/116/</guid><description>📢 This article was translated by gemini-3-flash-preview I thought about doing this back when I had my own server, but since I could just set up another site on the server to solve it, I never looked into it deeply. Now that my server has expired and I’ve moved my blog to GitHub, that’s no longer an option. So, I went looking for a solution.
The Failed Attempt Even though it failed, I’m documenting it anyway. I probably just didn’t know how to use it.</description></item><item><title>IP Rules for THE IDOLM@STER MILLION LIVE!</title><link>https://blog.yexca.net/en/archives/114/</link><pubDate>Fri, 09 Jun 2023 23:32:09 +0800</pubDate><guid>https://blog.yexca.net/en/archives/114/</guid><description>📢 This article was translated by gemini-3-flash-preview Recently, I couldn’t connect while playing Mirishita, so I looked for a solution. (Also testing a Spotify embed—this theme doesn’t support direct H5 parsing, and I’m too lazy to fix it.)
Domain 1 theaterdays.appspot.com Originally, a Spotify song by Amayadori was embedded here.
I’ve been loving Amayadori’s music lately and listening to them often. On Last.fm, “Nekokaburi” was my #1 track for the past 30 days. But the top …</description></item></channel></rss>