If you have more than one audio device, you’ve probably hit this. You’ve got headphones, speakers, a mic, maybe a monitor with its own audio. You plug something in or unplug it, or a driver updates, and Windows quietly decides that’s your new default.

I got tired of fixing it by hand, so I built SoundLatch.

You rank your audio devices, and SoundLatch makes sure Windows always uses the highest-ranked one that’s connected. My setup looks like this:

1. Headphones
2. Desktop speakers
3. Monitor audio

When I unplug my headphones, it switches to the speakers. When I plug them back in, it switches back. Microphones get their own separate list. If you just want one device to stay put, you can Lock it.

It’s a lightweight native app that sits in your tray, and it’s fixed a daily annoyance for me. A few dozen people have tested it so far and the feedback has been really positive.

Available on the Microsoft Store: https://soundlatch.formandelement.com/

PS: It's a few bucks, but if that’s a problem, DM me and I’ll sort you out with a copy!

Hi everyone,

I'm working on Uptime, a lightweight Windows desktop app designed to monitor system uptime, boot duration, and overall PC status.

I'm planning the next version and would like to hear directly from the community.

What features would you absolutely want to see in Uptime?

Here are a few areas I'm considering:

• Uptime history: detailed records of system uptime, restarts, and shutdowns.
• Advanced system monitoring: additional information about Windows health and system stability.
• Customizable dashboard: choose which statistics and system information to display.
• Notifications and alerts: get notified about unexpected restarts or other system events.
• Exportable reports: save uptime records and system statistics for later analysis.
• More customization: additional themes, interface options, and quality-of-life improvements.

These are just ideas, not a fixed roadmap. I'd rather prioritize features that people would genuinely find useful than add unnecessary complexity.

If you could choose just one feature for the next release, what would it be? And is there anything missing from existing Windows utilities that you would like Uptime to do better?

I'm especially interested in suggestions that keep the app lightweight, reliable, and easy to use.

Thanks for your feedback!

Hey everyone,

I just finished and released my first small desktop software project, and I'm mainly posting it here because I want feedback.

It's called Batch Renamer.

It's a very simple Windows utility for renaming multiple files at once.

The idea is deliberately boring:

1. Drop/select your files

2. Define a naming pattern

3. Preview the result

4. Rename them

5. The processed files are placed into a separate BR\_results folder

I wanted to keep it extremely lightweight, so it's portable. There's no installer, account system, subscription, ads, or cloud processing.

The application runs locally on the user's computer.

There is also no "AI" component collecting data or secretly sending files somewhere. The files being renamed stay on the local machine.

I built it because I wanted a tool that could just do one thing without turning into a giant application.

I'm not looking for customers or trying to sell anything here. I genuinely want to know whether this is actually useful to other people.

I'd especially appreciate criticism:

\- Is the UI understandable?

\- Is the workflow intuitive?

\- Would you actually use something like this?

\- Is there anything that feels unnecessary?

\- Is there something obvious that I'm missing?

\- Would the portable-folder approach bother you?

\- Is there anything about the way the files are handled that you would change?

If you think it's pointless, that's useful feedback too. I'd rather hear that now than spend another month building something nobody needs.

GitHub release: https://github.com/abdullahMTQ/batchrenamer

It's completely free at the moment.

Windows warning: Because the release is currently unsigned and uses a .bat launcher, Windows may show a security warning saying "Unknown Publisher." This is a Windows trust/signing warning, not an application error. You can inspect the files yourself before running them.

Thanks to anyone who takes a look. I'm genuinely interested in what you think.

The biggest change is under the hood: we’ve introduced an all-new native C++ TTS audio engine. By removing a lot of the Python overhead, we're seeing roughly 2x–5x faster generation depending on the model and hardware, along with lower memory usage and faster loading.

We’ve also spent quite a bit of time improving support for people who don’t have huge amounts of VRAM.

⚡ Native C++ TTS Engine

• Around 2x–5x faster generation
• Lower memory/VRAM overhead
• Around 30% faster app startup
• Native acceleration for supported models

🔴 AMD GPU Support

AMD users can now use native ROCm/HIP acceleration.

This has been one of the most requested Windows features for a while, so I'm particularly interested in hearing how it performs across different Radeon cards.

The video attached to this post shows it running on an RX 9060 XT.

🧠 Quantized Models

We've added Large and Medium quantized versions of larger TTS models including:

• Qwen3 TTS
• Chatterbox
• OmniVoice
• Vibevoice

This should make these models much more practical on GPUs with limited VRAM.

🎬 New AI Dubbing Studio

This is the other big part of the update.

You can now:

• Drop in a video or audio file
• Automatically transcribe it
• Detect up to 4 different speakers
• Translate the dialogue
• Rename/reassign speakers
• Assign a different AI voice to each speaker
• Generate synchronized dubbed speech
• Export the resulting dub

There’s also HTDemucs integration, so OpenVox can separate vocals from the original background audio and preserve things like music while replacing the dialogue.

The video demonstrates the full workflow by taking a video, identifying its speakers, separating the background track, translating the dialogue to Spanish, assigning voices and generating the final dub.

📝 Multi-Speaker Transcription

Drop & Transcribe can now perform speaker diarization as well.

You can rename speakers directly, reassign incorrectly detected segments and export a speaker-labelled, time-coded SRT.

We also fixed the Sortformer speaker diarization issue introduced in v2.3.0.

🔊 Select & Read Improvements

Select & Read has also received some attention in this release:

• Better reliability
• Smoother reading experience
• New draggable Select & Read overlay

So you can move the controls somewhere convenient instead of having them stuck in one position.

🔌 API

Pause tags are now supported through the API as well.

And there are a bunch of smaller UI optimizations and bug fixes throughout the app.

Would especially love feedback from AMD GPU users and anyone trying the new Dubbing Studio / multi-speaker transcription workflows.

If you run into something weird, let me know your GPU, VRAM and the model you were using. It helps a lot when tracking down hardware-specific issues.

Download:

https://openvoxai.com/

Hey everyone,

I've always wanted a file manager that's both blazingly fast and actually pleasant to look at and use every day. Explorer lacks basic power tools, and existing alternatives often feel bloated. So I built Sasha, powered by Tauri v2, Rust, and Svelte 5. AI was used heavily in the process, there's no reason to lie.

My goal was to make it incredibly snappy while bundling all the daily utilities you normally need 10 different background apps for.

Here is the quick rundown of what’s inside:

The Workflow: Dual-pane layout with tabs, 5 view styles, pinned files, smart folders, and color tags. It includes a Stash Queue (pick files from multiple folders and paste them all at once), Folder Sync, word-level File Diff, Flat View, and an Auto-organizer for messy folders. Plus, an integrated drop-down Terminal that automatically cds to the folder you're browsing.

Search & Performance: Sasha installs a dedicated, lightweight background service that indexes over 1M files in ~1 second. It powers the global search modal (Ctrl+F) with instant as-you-type results, accelerates whole-disk space analysis, and calculates folder sizes, while idling at just ~2 MB of RAM.

17 Built-in Viewers (Press Space or Shift+Space to pop out): Preview and edit almost anything natively. It handles images (decodes PSD/TIFF and includes a multi-layer editor), videos (includes a timeline editor to trim/export clips), an audio capsule player in the status bar, Office docs, spreadsheets, PDFs, EPUBs, Comics, Code/MD, 3D models, SQLite DBs, GPS tracks, Fonts, and SSL certs. You can also peek inside archives or 1-click mount disc images.

Power Utilities: Disk space analyzer, on-the-fly SHA-256 checker, batch renamer, batch image resizer, and dedicated batch converters for audio, video, and images.

Customization: Theme selector with 21 built-in presets, a visual Theme Studio (supporting custom image wallpapers, live looping video, and interactive web/shader backgrounds), the community Theme Hub, and 10 languages.

Trial: 7 days unrestricted. Once expired, nothing is disabled; you can keep using it freely with a WinRAR-style pop-up unless you decide to buy a license.

It’s available today for Windows 10/11 (Linux build is coming soon).

The first public build (v0.1.0) is up for download here: Github

I'd love to get your honest feedback, bug reports, or feature requests.

Separate notes for separate projects just makes sense to me.

What never made sense was having to remember where I put the note.

So I built ClingNote.

You can attach a note to a file or app, and when you switch what you’re working on, ClingNote switches to the matching note automatically.

Spreadsheet → budget notes.
Power BI report → project notes.
PDF → review notes.

You can still use normal standalone notes too, with rich text, markdown, tables, images, code blocks, deep search, etc.

Next up is Tasks (checklists) with the same idea use them normally, or attach them to a file so the checklist follows the work too, integrated perfectly with your notes.

It’s on the Microsoft Store now.

Does anyone else actually want notes organized around the thing you’re working on, instead of folders/notebooks?

I had always wanted a modern Windows terminal that emulated a CRT monitor—something like Cool Retro Term for Linux. However, nothing of the sort existed for Windows, aside from a rather pathetic attempt at simulating scanlines in Microsoft Terminal. Even when Cool-Retro was eventually ported to Windows, the quality of the effects in the port seemed inferior to the Linux version.

At the time, I was developing a game and its custom engine -- featuring a highly realistic, deeply customizable CRT simulation (inspired by Cool-Retro) -- because the game was about hackers and geeks in 1980s, requiring interaction with a wide variety of computers and terminals. Then it hit me: I could take that shader from the game, combine it with the native Windows ConPTY, and easily create a Cool Retro Term equivalent -- one that was even more realistic and customizable, and ran natively on Windows. The first prototype was ready in just a couple of days, and it worked beautifully! Of course, it required a lot of polishing, and I also wanted to implement several other useful features, such as:

• Multi-tab support with individual style settings.
• Console buffer search, quick text copying via the middle mouse button, etc.
• Smooth, pixel-perfect vertical scrolling- both in the shell and in applications like Vim.
• A built-in mini web browser optimized for keyboard control. Useful for keeping documentation or a web service right inside the terminal.
• A mode for viewing all terminal sessions as thumbnails - "Win-Tab" style.
• A GUI settings panel (you can also just simply edit the JSON config).
• The ability to open new tabs with specific CRT-theme directly from the command line.
• A toggleable AI assistant panel based on Codex that can interact with the console on your behalf or collaborate with you.

Regarding that last feature -- I realize people often look askance at apps that integrate with generative LLMs. But let’s be honest: these days, AI is being shoehorned into almost every piece of software, whether it makes sense or not. But in this specific case, it is more than justified. I believe it’s incredibly useful to be able to ask the AI in plain English to turn a request into a PowerShell command, help configure Vim, construct a complex SQL query, and so on, right within the terminal. The key advantage here is that you see everything it writes to the console, allowing you to learn new commands and tricks. It embodies the IBM principle: the machine should do the work, while the human makes the decisions. And if you’re an expert, you can simply disable the panel.

I’ll admit it: Scanline Term itself was created with some AI assistance. It wasn't exactly "vibe coding," but rather agent-based development where I handled the architecture and high-level functions, while the AI ​​did the grunt work. Nevertheless, it is a native, polished, and compact application -- clocking in at around 19 MB, it launches instantly, is memory-efficient, and runs fast.

By the way, the CRT simulation has a strange psychological effect - at least for me. It boosts concentration and focus on the text. When it’s running in full-screen mode, glowing and "breathing" with that analog feel, it somehow draws me in and helps me concentrate. While I used to work in the console only sporadically, now I actually want to jump in and immerse myself in that cozy phosphor glow -- just me and the code, the text, or a technical process. I experience something similar with vintage tube amplifiers and speakers; their unique sound draws me in and shuts out the outside world - a world that has hooked us all on quick dopamine hits.

In any case, this product turned out to be unexpectedly useful to me. I’d be happy if it proves helpful to others, too, and I’m always grateful for any feedback.

https://github.com/z-hunter/Scanline-Term