By giving each voice its own section and letting them meet only in the chorus. The release of Teresa & Maria runs 2:59 (Wikipedia). Jerry Heil carries the sung opening and the bridge, alyona alyona carries the rapped middle in Ukrainian, and the chorus belongs to both of them, with its last line in English. The music is credited to Anton Chilibi and Ivan Klymenko, the lyrics to the two performers. A rapper and a singer on one track is an arrangement problem first and a mix problem second, and getting the order right decides most of what follows.
Who sings what, and why does the order matter?
The form runs sung verse, pre-chorus, double chorus, rapped verse, sung bridge, double chorus. The listener meets the singer first, hears the hook twice, and only then gets the rapper, so the rap arrives as a change of texture inside a song whose identity is already set. Open with the rap instead and a jury has filed the song under one genre before the hook has played once.
The lyric sheet shows the second device in that opening: the first word of each line is broken into repeated syllables before it resolves. A sung line becomes a rhythmic figure, so the singer's section already carries the pulse the rapper will take over later. By the time alyona alyona enters, the ear has been prepared for a voice made entirely of rhythm. NPR's Glen Weldon, previewing the final, wrote that "the two performers remain distinct" (NPR), which is the arrangement doing its job: the contrast is the point, and nothing in the mix should smooth it away.
What does a rapped verse need from the mix that a sung chorus does not?
Level control and consonants. A rapped line in Ukrainian sits in a narrow dynamic window and carries its intelligibility in consonants that live roughly between 2 and 5 kHz and arrive as short bursts. A sung chorus carries its identity in sustained vowels, with the weight lower, between 200 Hz and 1 kHz, and long enough for a reverb tail to be heard on every note. Run both through one vocal chain and one of them loses.
The working split is two chains. On the rapped section: compression with a fast attack, 1 to 3 ms, around 4:1, taking 3 to 6 dB off the loudest words so the consonant peaks stay inside the beat; a short space, a room under a second or a slap below 100 ms; a small presence lift near 3 kHz if the drums are dense. On the sung sections: a slower attack of 10 to 30 ms so each vowel keeps its onset, a gentler ratio, and a longer space with 40 to 80 ms of pre-delay so the words stay in front of the tail. My note on serial compression covers why the sung chain often ends up as two light stages instead of one heavy one.
How do you keep the handoffs from sounding like a level jump?
Match loudness, ignore peaks. A rap vocal peaks lower on a meter than a belted chorus and still reads louder to the ear, because its energy is concentrated in the 1 to 4 kHz band the ear weights most heavily. Set the two sections so their short-term loudness, measured over the 3-second LUFS-S window, lands within about 1 LU of each other across the handoff bars. The peak meter will show a 4 to 6 dB difference between the sections and will be right about that too. The listener hears continuity.
The second fix is spatial. The rapped verse gets its own space and the sung sections get theirs, but in the choruses, where both voices sing, they go into one shared reverb. Two voices in two different rooms sound like an edit. Two voices in one room sound like a duet, which is what the chorus is.
What does a bilingual chorus do to a de-esser?
The chorus switches from Ukrainian to English inside a single section. Ukrainian sibilants and affricates, the sounds written ш, щ, ч and ц, do not sit in the same band as an English s, and they last longer. A de-esser tuned to one language's s either misses the other or takes half a word with it. The answer is two detection bands, either a dynamic EQ with two nodes or two de-esser instances keyed to different frequencies, and a check of the chorus line by line rather than a single setting for the whole vocal.
Why does 2:59 matter?
Eurovision caps a performance at three minutes, and Ukraine's Vidbir applies the same limit. Teresa & Maria came out at 2:59 on 11 January 2024, one day before the Vidbir 2024 field was formally announced. The song was written at the length rather than trimmed to it afterwards. A sung verse, a pre-chorus, two double choruses, a rapped verse and a bridge inside 179 seconds leaves no room for an instrumental intro or a repeated outro, and the record shows it: the voice enters almost at once and the final chorus is the ending. My earlier note on what changes in a contest version describes the 30 to 50 seconds most releases have to lose. This one had already lost them on paper.
How did the staging use the same split?
Tanu Muino, the music video director appointed creative director for Malmö, lit the two singers differently: Heil in yellow-orange, alyona alyona in cool blue, the colours of the Ukrainian flag, and the two met for the final duet (Wikipedia). That is the arrangement drawn in light. The structure of one voice, then the other, then both was legible enough that a staging team could build on it without a single explanatory line.
What did the scoring say about the structure?
Vidbir 2024, announced on 4 February after a one-day delay: 10 points from the jury and 11 from the televote, the maximum available. Malmö, first semi-final: second, with 173 points. Grand final: third, 453 points, split 146 from the juries and 307 from the televote, with two jury sets of 12 (Czechia and Moldova) and seven televote sets of 12. The televote share, 68 percent of the total, is the figure a writer looks at, because the public votes on memory and a two-voice structure gives memory two handles. The chart side of the story is in a separate note. Full credit list with sources at /credits.
Frequently asked
Who wrote Teresa & Maria? The music is credited to Anton Chilibi and Ivan Klymenko, the lyrics to Aliona Savranenko (alyona alyona) and Yana Shemaieva (Jerry Heil). It was released on 11 January 2024 through Enko.
How long is Teresa & Maria? The release runs 2:59, already under the three-minute Eurovision limit, so the contest version never needed an edit for length.
How did Teresa & Maria score at Eurovision 2024? Second in the first semi-final with 173 points, then third in the grand final with 453 points, split 146 from the juries and 307 from the televote.
If you are writing for two voices
Settle the form before the beat: who opens, who owns the hook, where the two meet. Then build two vocal chains and one shared space for the sections they sing together, and check the handoffs on a loudness meter rather than a peak meter. Background and contact at /about.