v2.5.0 just published
new feature:

1. TTML lyrics with translate
2. Using asynchronous download scheduling doubles the speed of batch downloads.
3. Accelerate Dolby Atmos format encoding and decoding
4. add “new version” detect feature (This won’t work until my GitHub account is unbanned）

amdl cli also optimizedget-lyrics logic and add more and smarter parameter degradation strategies

rewrote the readme file

For the past few weeks, I’ve been pouring my heart into building AMLL PiP Player—and honestly, I’m beyond proud of how it turned out! I’ve always been obsessed with Apple Music’s gorgeous, bouncing syllable-by-syllable lyrics and colorful ambient fluid glow, but I hated that there wasn’t a lightweight, cross-platform way to have that pinned on my desktop while coding or doing homework. So instead of waiting for someone else to make it, I built it myself completely from scratch in the browser using Vue 3 and TypeScript! By taking advantage of the cutting-edge Document Picture-in-Picture API and File System Access API, I managed to create a 100% client-side player that indexes your local tracks, automatically pairs them with rich .ttml / .lrc lyrics and album art, and pops out into a floating, frosted-glass desktop card that stays pinned on top of any window. Pushing the limits of what a high schooler can build with modern web standards, no backend, and zero data tracking—it feels surreal seeing an aesthetic I dreamed of running so smoothly on my own screen!

This project is designed specifically for individuals with ADHD. Research indicates that the ADHD population faces physiologically unique and severe cognitive challenges:

First, their working memory bandwidth—responsible for holding fleeting thoughts—is extremely narrow and exceptionally prone to collapse. While a neurotypical brain typically retains working memory for 10 to 15 seconds, an ADHD brain often loses it in just a few seconds. Abnormalities in prefrontal dopamine pathways render their attention extraordinarily fragile. Any minor external stimulus—such as unlocking a screen, a 0.5-second app cold-start delay, or cumbersome “New Note” buttons and categorization workflows—acts like a disruptive sensory intrusion. It instantly shatters and dissolves newly formed neural circuits, causing brilliant sparks of inspiration to vanish forever. What is most devastating is knowing they just had a brilliant idea, yet it has been permanently erased. This “acute thought evaporation” frequently triggers deep self-blame and mental exhaustion.

Second, they struggle with severe executive dysfunction and “initiation paralysis.” The vast majority of productivity and note-taking apps on the market are built around neurotypical prefrontal logic, forcing users through preliminary steps like titling, choosing folders, manual tagging, and formatting before they can even begin recording. While simple to neurotypicals, these prerequisites are instantly amplified in the ADHD brain into an overwhelming, suffocating decision-making load, inevitably leading to executive overload and abandonment of the task.

Furthermore, individuals with ADHD suffer profoundly from “Object Permanence Failure” and “Retrieval Cue Failure.” Under the premise of “out of sight, out of mind,” content buried within nested folders quickly turns into a permanently forgotten “digital graveyard.” Conversely, keeping everything on the home screen triggers intense visual overload and avoidance anxiety. When attempting retrieval, they often struggle to recall exact keywords, rendering traditional keyword-dependent search completely useless.

This is precisely why Fleeting abandons the conventional, counterproductive trap of feature bloat. Instead, it embraces neurodiversity through radical, physics-level reduction. By compressing the entire process into a single, intuitive “Launch-to-Record” action and silently attaching implicit spatio-temporal fingerprints (such as time and context) in the background, Fleeting shifts the burden from “active keyword recall” to “contextual reconstruction and visual recognition.” It serves as a secure, guilt-free external thought airbag for these fragile yet precious sparks of insight.

I want to add English support first. And then I give up

InspireFlow

An AI-powered creative workflow assistant for creators.

InspireFlow (升变) is a project built by a team of high school students during AdventureX 2026. It explores how AI can help creators manage the journey from an initial idea to a finished piece of content, while keeping their creative assets organized and under their control.

What are we building?

Creating content often involves scattered ideas, unfinished drafts, collaboration messages, brand requirements, and production plans across different tools.

InspireFlow aims to provide a unified workflow where creators can:

• Capture and organize creative ideas
• Transform rough thoughts into structured content plans
• Assist with outlines, scripts, and production workflows using AI
• Keep track of creative assets and collaboration history
• Create verifiable records for important creative materials

The goal is to make the creative process more organized, transparent, and easier to manage.

Why InspireFlow?

Many creators spend significant time developing ideas, writing drafts, and communicating with collaborators. However, these valuable creative assets are often stored across disconnected platforms, making it difficult to track ownership, changes, and the evolution of an idea.

InspireFlow explores a different approach: creators should have better visibility and control over the creative process behind their work.

How it works

InspireFlow combines several technologies:

• AI assistance — helping creators structure ideas and improve workflows
• On-device processing — designed to keep sensitive creative information local whenever possible
• iOS application — providing a native experience for creators
• Hardware interaction — integrating with devices such as smart rings for easier idea capture
• Blockchain-based records — exploring ways to create verifiable records for creative assets

Current Status

InspireFlow is currently a prototype/demo developed during AdventureX 2026.

The project was built within a short hackathon timeframe, including:

• A complete iOS demo application
• AI-assisted creative workflows
• Hardware integration experiments
• A prototype for creative asset verification

There is still a lot to improve, and we are continuing to explore how AI can better support creators.

Vision

From the first spark of an idea to the final delivery, creators should have a clearer and more connected workflow.

InspireFlow is an experiment toward building tools that help creators focus on creativity while maintaining better control over their creative process.

I’ve been working on the AppleMusic-Downloader project, primarily debugging a Windows-specific issue where the desktop app would hang during downloads. The root cause turned out to be PyInstaller’s --windowed mode setting sys.stdout and sys.stderr to None, which broke async libraries like httpx and asyncio. The fix was calling AllocConsole() on Windows to allocate a visible console window and properly redirect all three standard streams (stdin, stdout, stderr). I also cleaned up unnecessary debug code (log file output, /api/logs endpoint), added step-by-step logging in the download pipeline to help diagnose future issues, fixed a few i18n key mismatches in the frontend, and added a note in both READMEs telling Windows users not to close the console window.

The Dolby format finally downloaded normally. During this stage, copilot helped me debug and test the project.This project is ready to ship

published to PyPi!

workflow currently running!!!

All the rebuild work is complete; it’s time to open the PR.

The interface is ready to use!

I felt this didn’t quite meet my aesthetic standards for the sidebar, so I decided to use Figma to design the UI.

Successful integration of the front-end and back-end.

Remove Windows-specific GUI components and associated files from the project, including CMake configurations, resource files, and source code for the Flutter window implementation. This cleanup streamlines the project structure and eliminates unnecessary dependencies for non-Windows platforms.