How to turn a YouTube cocktail tutorial into a recipe card you can trust
You learned the drink from a video, and now you want the spec written down somewhere you can read it with a shaker in one hand. My shortcut is to paste the link into a youtube video transcript generator, pull the words the host actually said, and rebuild them into a card: ingredients, measures, method, glass, garnish. The text gets you most of the way there. The rest is where the drink lives or dies, because captions are bad at hearing numbers and even worse at hearing liqueur names.
This works for a home bar with a dozen bottles and for a crew learning a new menu before Friday. The process is the same. Only the binder gets thicker.
Paste the watch link, a youtu.be share link or a Shorts link (plenty of quick cocktail clips live in Shorts) and hit GET TRANSCRIPT. The tool is free, asks for no account and needs no browser extension. It reads the caption tracks YouTube already has for that video, so before you download anything, open the "Transcript language" menu. If you see both "English" and "English (auto)", pick plain "English": that track was uploaded by the creator or a captioner, while the auto track is YouTube's speech recognition guessing at what was said over a rattling tin.
Leave the "Timestamps" switch on and download the .txt. Lines come out like [0:27] Good morning..., and that minute mark is your map back to the pour. You will need it.
Some limits first. If the video has no captions at all, you get a message saying no subtitles or transcript are available, and you are back to pausing and scribbling. Private and members-only videos won't load either, and a tutorial uploaded an hour ago may not have captions yet. Some lines keep the caption's mid-line breaks, so expect a minute of tidying. The page does work at phone width, handy when your laptop is nowhere near the bar.
Every number and every proper noun in the transcript is a suspect until you have seen it on screen.
Measures are the dangerous part. "Half an ounce" and "an ounce and a half" sound close at speed, and if "and a half" gets swallowed under the sound of ice, your sweetener just tripled or vanished. "Three quarters" can come out as something that isn't a measure at all. "Barspoon" splits into two words, "dash" and "dashes" blur together, and "top with soda" sometimes loses the soda.
Names are the second trap. Orgeat, falernum, Cointreau, Chartreuse, Lillet, maraschino: none of these are words a general speech model expects, so they get swapped for whatever English phrase sounds nearest. A wrong brand can mean a different proof or sugar level.
The fix is boring and it works. For each ingredient line, jump to its timestamp in the player and watch the jigger. If the host says one amount and pours another, write down what went into the tin and add a note. Rewatching also shows what the audio never says, like which side of the jigger they used or how long they really shook.
One US fluid ounce is about 29.6 ml, and most bartenders round it to 30. That makes 2 oz roughly 60 ml and 3/4 oz roughly 22.5 ml, which nobody can pour, so it gets written as 22 or 23. The problem comes when a creator works in metric with 25 ml and 50 ml jiggers and you convert their numbers to ounces and back again. Each round trip shaves or adds a little, and the balance drifts.
My rule: write the measure in the unit the host used, put the conversion in brackets after it, and then check that the ratio still holds. If the spirit-to-citrus-to-sugar ratio changed during conversion, you made the mistake, not the bartender.
Here is a generic classic Daiquiri, a spec found in many places and credited to nobody in particular. The layout is what matters.
Daiquiri
Source: classic spec (for a video card: creator, channel, video title, timestamp of the pour)
Glass: coupe, chilled
Ice: shake with cubes, serve up
2 oz (60 ml) white rum
1 oz (30 ml) fresh lime juice
3/4 oz (22 ml) simple syrup, 1:1
Method: shake hard, double strain into the chilled coupe
Garnish: lime wheel, or none
Notes: checked against the pour at [timestamp]; my changes go here
People skip the source line, then six months later can't remember why the card says 1/2 oz of something. Put the timestamp of the pour on it as well as the video title.
One card per page, in sheet protectors, grouped by family (sours, stirred, highballs) rather than by base spirit, because that is how you think when a guest says "something like a margarita but not a margarita." Keep the downloaded .txt in a folder with the same name as the card. Each transcript also gets its own page link, and reopening it loads the text again, so you can paste that link into the card's source line instead of hunting for the video later.
For bar staff, add one more field: house spec. The video's version and your bar's version will drift apart (different syrup, different ice, a manager's preference), and a new hire needs to see which one is the rule.
Pull both transcripts and line them up; the differences jump out faster than from watching two videos back to back. An illustration for a sour:
|
Line |
Video A |
Video B |
What to taste for |
|
Sugar |
3/4 oz simple syrup, 1:1 |
1/2 oz rich syrup, 2:1 |
Similar sweetness, different texture |
|
Citrus |
1 oz lime |
3/4 oz lime |
Is the second one flat or just rounder? |
|
Method |
Hard shake |
Shake, then double strain |
Ice chips on the surface or not |
|
Garnish |
None |
Lime wheel |
Aroma on the first sip |
Check every cell at its timestamp before you start mixing, because a misheard "half" in one column turns a fair comparison into a broken drink. Then make both side by side, half-size if you're tasting alone.
A list of ingredients is hard for anyone to own, but the host's words are their work: the method notes, the jokes, the story about where the drink came from. Your private binder can hold whatever helps you learn. Once a card goes anywhere public, like a menu blurb, a blog post or an Instagram caption, rewrite the method in your own words and name the creator and the video. Never paste a cleaned-up transcript and call it your recipe.
Last tip from my own binder: I date the "checked against the pour" note. Creators reshoot, and the date tells me which video I checked.