Four lines sung back to you, twenty credits an attempt

Each generated track is 20 credits regardless of length. Ten attempts is 200 credits. A $15 top-up of 1,000 credits buys fifty.

That price makes the correct process affordable. Stop describing your way to a perfect result on paper. Generate a batch and listen.

You supply two inputs. The lyric, and a plain-language description of the sound. Bright, simple, upbeat, a little retro, one voice. More detail than that mostly buys a busier track that gets in the way of the words.

Judge them the way they will be encountered. Play a candidate once, go and do something else for ten minutes, then see which one comes back unprompted.

The one you can still hum is the one, even when it is not the one you preferred while listening. Those two are different tracks more often than anybody expects.

A jingle exists to survive the gap between hearing something and needing it. That gap is usually weeks, and almost nothing crosses it.

Not your positioning. Not your differentiator. Not the tagline the agency was pleased with.

So the test is not whether it is enjoyable. The test is whether it comes back uninvited.

The mechanisms are old and slightly embarrassing. A short melodic phrase. Heavy repetition. A rhyme that closes cleanly. The brand name landing on the strongest beat.

Count the jingles you can hum for brands you have never bought from. Then count the taglines you can quote for brands you buy weekly. The scoreboard is not close.

So the brief to yourself is short. Make it repeatable rather than enjoyable, and stop trying to make it sound expensive.

Keep it to four lines or fewer. Put the brand name at the end of a line so the rhyme lands on it.

Say what you do in plain words rather than a positioning statement. People remember we deliver on Sunday. Nobody has ever remembered reimagining convenience.

Repeat something. The same phrase twice inside eight seconds feels excessive while you write it, and correct the moment you hear it sung.

Use words that are easy to sing. Long vowels carry. Hard consonant clusters do not. If your brand name is awkward, spell it phonetically so it is sung the way customers say it.

Almost every failed jingle fails the same way. The name sits in the second line, unstressed, competing with an adjective. The client signs it off because the music sounded expensive.

Read the four lines aloud to somebody before you spend a credit. If they cannot repeat the last line back to you, the lyric is not finished.

A jingle is not a bed. A bed runs under the whole read at a low level and you should stop noticing it. A jingle is an event, and events belong at the ends.

Two shapes work. Sing it over the first two seconds, before the read starts, so the name arrives before anything is being sold. Or sing it at the tail, over the last line, as the thing left behind.

Both fit inside a thirty second ad without stealing time from the offer. Four sung lines take about six seconds. That is a fifth of the ad, and about as much as you should spend.

Set the level higher than you would set a bed. A jingle you have to strain to hear has failed at its only job.

And test the ad without it. Export both, watch them on a phone, and keep the one you would sit through twice.

The words have to come from you. From somebody who knows the business, the customer, and how the name sounds in a mouth. That is where a jingle is won.

It does not sync to your edit. There is no way to make the hook land on a chosen cut, so treat the jingle as something that plays over a section.

It is not revised. You cannot keep the melody and fix the third line. You generate again with different words, which at 20 credits beats a revision round.

It will not imitate an existing jingle or a named artist, and that is the point. The result would be somebody else's asset rather than yours.

And a sung claim is not a trusted claim. A medical result reads as advertising the moment it is sung. Film a person saying that one plainly and run the take through the same editor for 100 credits.