Under maintenance
ToolHub
Streamline Tools, Master Productivity
No need for keywords — just describe the task and I'll find the right tool on this site.
Hold to talk and the AI instantly turns your speech into fluent text (polishing) or a translation; inference runs entirely on-premise, and no data leaves the internal network.
Under maintenance
AI image generation from text and images: Krea 2 Turbo text-to-image, RealVisXL image-to-image and masked local editing; prompt optimization preserves the user's full intent.
Record one Chinese and one English clip to create your voice; the AI then reads your script in your timbre (Chinese answers sound like you too), sentence by sentence with highlighting that shows where it is, and pausing and resuming does not recompute. Voiceprints are stored only in your browser by default; you can choose to store them on the server after giving explicit consent.
Under maintenance
Generate a video from one image plus a description in Chinese, or take the last frame of the previous clip and extend it (MiniMax H3 on DGX Spark): Chinese prompts work directly without translation, 864×480 / 5 seconds takes about 80 seconds, with a native 32 kHz audio track; an asynchronous queue with live progress, true cancellation, and automatic polling resume after a refresh.
A ChatGPT-style chat tool wrapping the DGX Spark LLM, with streaming replies, collapsible reasoning, image input (vision), and a code Canvas with live preview and editing.
Generates short BGM loops and sound-effect assets locally on DGX Spark (Stable Audio Open). Enter a Chinese or English description, let AI optimize the prompt, and pick a style and duration (including an 11-second quick version).
An experimental tool on local DGX Spark that turns lyrics into vocal songs with YuE. A Suno-style test panel supporting length presets from 30 seconds to 5 minutes, Chinese/English lyrics, genre presets and prompt optimization.
Enter a description, have AI generate an image suited to be an icon, and convert it at high quality into a complete favicon package (multi-size .ico, PNGs, apple-touch icon, manifest and HTML snippet); supports a letter-mark mode and live tab preview. Runs entirely on-premise.
A handwritten diary that writes back. Write any time; once the ink is absorbed, Tom slowly appears word by word (in Kaiti font) on the line below yours. On-premise AI; supports fullscreen.
Under maintenance
Upload a reference image, write an action or plot description, and choose the target duration; the AI automatically breaks it into multiple coherent scenes, generates them one by one, and stitches them seamlessly into a complete short film (with sound).