ChatGPT 对话自动整理到 Obsidian 日记 (LM generated — please review and edit before saving)
Paste a long-video or podcast transcript and get a clip map: the 10 strongest self-contained moments to cut into shorts, each with the exact quoted start and end lines, a one-line reason it works as a standalone clip, a context-risk flag, and a working title. It only selects moments from the text you paste. It does not write scripts, edit video, or publish. For hooks and scripts for a chosen moment, Hook Lab is the next step.
Turn a script into a vertical narrated short with voice and subtitles that light up word by word. Speech recognition recovers the timings from the generated audio, so captions never drift when you change the voice, the speed, or a line.
YouTube VTT & podcast transcripts → structured PKM notes: TL;DR, key points, quotes, tags, batch index. One file or folder. Any AI agent. No API keys.
Use when your Japanese draft still reads AI-written. For bloggers, PR teams, marketers, and students: paste the text and get a human-sounding rewrite, plus a diff of every AI tell removed. It replaces vague AI phrasing — 浮き彫りになった, 多面的な, 〜ではないでしょうか, 今後の展開が期待される — with concrete facts and one clear voice. Facts are never invented or changed, only the wording. Output = the rewritten full text + a change log of what was fixed.
Not Google Translate. Not "good enough." This preserves a character's voice, catches cultural landmines before launch, generates titles that actually perform, translates the onomatopoeia most translators give up on, and adapts content across markets — built by a native Japanese speaker who ships to both Japan and overseas. Paste 5 lines of a character and watch it lock their voice forever. One question routes you to the right mode in seconds.
Use when your academic or research writing reads AI-generated. For students, researchers, and writers: paste your draft, get a version that reads like a careful human author — inflated symbolism, promotional language, em-dash overuse, rule-of-three, vague attributions, and AI vocabulary stripped out — argument and citations intact. Based on Wikipedia's 'Signs of AI writing'. Output = rewrite + a list of every AI tell fixed.
Compresses AI filler text — preambles, transitions, repetitions — at 3 intensity levels (Lite / Full / Ultra) while enforcing a protected scope (URLs, numbers, code, reasoning) that is never deleted, so "be concise" stops silently destroying what matters.
Humanizer from NousResearch’s Hermes project, backed by the 228K+ star NousResearch GitHub ecosystem. Use it to remove common AI writing patterns, generic filler, and awkward phrasing while preserving the original meaning — making blogs, emails, essays, posts, and other text sound more natural and human.