Skip to content

hubeiqiao/how-i-make-videos-with-ai / skills/joe-speaking-talk-cut

joe-speaking-talk-cut

Verified

Use when cutting a recording of your own talk or demo at an event into a published video with a cover hook, captions, cards, feed covers and copy in your own voice.

FileKindSize
SKILL.mdskill md2.8 KB
examples/input-brief.mdmarkdown1.1 KB
references/workflow.mdmarkdown2.4 KB

references/workflow.md

View on GitHub
# Talk recording to published cut
Portable teaching adaptation of Joe's joe-speaking-talk-cut workflow. Not the original project, its scripts or its brand assets.

Inputs: stage recording, speaker photos, product screens, the speaker's post draft, destinations.
1. Make a working copy at a constant frame rate; convert HDR phone footage to SDR. Keep the original unchanged.
2. Transcribe with word timestamps and check against the audio. The first word of a phrase came out 0.3 to 0.5 s early, so cut where the voice really starts. A long pass once skipped a whole phrase: re-transcribe gaps on their own. Flag disputed words to the speaker.
3. Keep the self-introduction, the pitch, the whole demo, the result and reaction, and the closing lines. Captions keep the speaker's grammar.
4. Cut in measured silence with short fades. To remove an "um" inside a kept line, split in the silences around it and give the new piece its own framing; a cut on the same framing reads as a jump.
5. Check every speaker photo first: several had the speaker's eyes half closed.
6. Time each card to the words it illustrates. Caption pages: at most 7 words or 4.2 s, breaking at sentence ends, speaker changes and pauses over 0.9 s. Speaker labels show only while that person talks.
7. Sound: one constant gain per speaker turn, ramping only in the gaps, because venue speakers and off-mic voices came in 12 to 20 dB under the handheld mic. No music under the demo. Shortlist closing tracks by measured energy, let the speaker pick from two previews, and land the hit on a phrase start. For an A/B, render the picture once and swap only the audio. Joe found synthesized whooshes cheap; let the music's beats carry the hits.
8. Translated version: same edit and audio in a separate file, with a translated line under each original caption. Never overwrite the first-language final.
9. Covers: 16:9 frame 0 for YouTube and LinkedIn; 4:3 for a landscape video on RedNote, which crops covers to 4:3; 9:16 for Reels and TikTok, with the design inside the middle 3:4 band the profile grid shows (a 3:4 upload lost both sides). Check eyes, text and hands in each.
10. Copy: one file that opens with a table of platform, video and cover. Check every descriptive word (live, uncut, first) against the edit. On RedNote: no links, and none of the superlatives it flags.
11. Deliver HD only. Move older renders to the Trash instead of deleting them.
slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."Stars: 198kUpdated Claude CodeCodexVerifiedInstall
ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt.Stars: 196kUpdated Claude CodeCodexVerifiedInstall