Hey Product Hunt 👋 I built SquishyFile because every video tool I needed — upscaler, transcriber, GIF converter — made me upload the file to a server first. Slow, and honestly uncomfortable when the footage isn't something I want sitting on someone else's disk. So everything in SquishyFile runs entirely on your device, using WebAssembly and WebGPU: - AI Video Upscaler — for sources 720p and under, it runs a real neural network locally to reconstruct detail, not just stretch pixels. Larger sources fall back to a fast FSR shader instead, since a full AI model would be too slow in a browser at that size. Pick 2x or 4x, drag a before/after slider to check the result. - AI Transcript — an on-device speech-to-text model (Moonshine) transcribes your video's audio locally and exports TXT, SRT or VTT. Supports English, Spanish, Japanese, Chinese and Korean. The model downloads once and runs fully in your browser — no audio ever leaves your device. - MP4 to GIF — trim, crop, adjust frame rate/size/color count, change speed, even reverse the clip. No watermark, no file size cap. All three (plus a video compressor, MP4↔MP3, filters and a frame extractor) are part of a free toolkit I've been building solo, no paywalls, no pro tier holding features hostage. Would love feedback, especially on the upscaler and transcript tools since they're the newest and most experimental parts of the stack. Happy to answer anything about how the on-device AI/WASM side works.

Squishyfile

AI upscale, AI transcript, GIF. No watermark, no usage cap!

AI upscale, AI transcript, GIF. No watermark, no usage cap!

SquishyFile

Previous Squishyfile Launches

Previous Squishyfile Launches