Short answer. A traditional Ukrainian open-throat vocal peaks around 2.8 to 3.2 kHz and swings 20 dB or more between an ornament and a sustained call. A standard pop chain (bright large-diaphragm condenser, one fast compressor, a 3 kHz presence lift) doubles down on exactly the band that is already loudest and squashes exactly the dynamics that make the take sound human. The fix is boring and reliable: darker capsule, more headroom, two compressors doing small amounts of different work, and a carve in the arrangement rather than in the voice.
What actually makes an open-throat vocal different from a pop vocal?
Two measurable things.
The first is where the energy sits. The singer's formant, the resonance cluster that lets an unamplified voice cut through an orchestra, lives between roughly 2.8 and 3.2 kHz, and it was described in the acoustics literature decades before anyone put a compressor on it (see Sundberg's work on the singing voice). A trained open-throat singer produces it deliberately and constantly. A pop vocal produced close to the mic does not depend on it, because the microphone does that job instead.
The second is dynamic range. Village-style singing is built on the contrast between a quiet ornamented phrase and a full-chest sustained note. Measure a raw take and you will routinely see 20 to 26 dB between those two states inside a single verse. A pop vocal tracked at conversational distance sits inside 8 to 12 dB.
Everything below follows from those two numbers.
Which microphone and preamp survive that energy?
Start by not making it worse. A modern bright large-diaphragm condenser with a designed presence rise in the 3 to 8 kHz region is adding gain to a band the singer is already pushing. On a strong take that stacks into harshness before you have touched a plugin.
Three options that work, in order of how often they get used:
- A dark large-diaphragm condenser, tube or transformer coupled, with a flat or gently falling response above 2 kHz. You keep the detail and the air without the 3 kHz stack.
- A ribbon, when the room is good and the singer is loud. The natural high-frequency roll-off does the carve for free, and modern active ribbons have enough output that preamp noise stops being the problem it used to be.
- A large-diaphragm dynamic when the take is very loud or the room is not treated. Less flattering, far more forgiving.
Set the pad before you set the gain. A 10 dB pad on the capsule costs nothing on a loud source and keeps the preamp out of its own distortion region, which on most transformer designs starts producing audible third-harmonic content well before the converter clips.
Aim for peaks around -12 dBFS, not -6. Modern 24-bit converters have no meaningful noise penalty at that level, and you leave room for the ornament that lands 8 dB above everything else in take four.
How do you compress it without flattening the ornaments?
One compressor cannot do both jobs. Level-riding across a phrase and catching a 3 ms transient are different time constants, and asking a single unit to do both is what produces the pumped, lifeless vocal that people then try to fix with saturation.
Split it:
1. Slow and gentle first. An optical design, roughly 10 ms attack, program-dependent release, 3 to 4 dB of reduction on the loudest phrases. This is macro level, verse to chorus, not transient control. Opticals are slow enough that ornaments pass through untouched. 2. Fast and small second. A FET design at 4:1. The 1176 attack control spans roughly 20 to 800 microseconds, and the fastest setting is fast enough to shave consonant peaks without audibly ducking the note behind them. 3 to 5 dB on peaks only.
Total reduction across both stages lands around 8 to 10 dB. The take still breathes because no single stage is working hard.
If you need more density than that, use parallel compression instead of pushing either stage further. Send to a bus, crush it at 10:1 with 15 dB of reduction, and blend it back under the dry vocal at -12 to -18 dB. You get weight without losing the top of the dynamic range.
What EQ moves keep a voice sitting next to a sopilka?
This is where most folk-pop hybrids fall apart, and the mistake is almost always the same: engineers carve the voice to make room for the flute, then carve the flute to make room for the voice, and end up with two thin sources and a hole in the middle of the mix.
Pick one. Usually the voice wins, because it carries the lyric.
A soprano sopilka's fundamentals sit above 500 Hz, which means its energy lands on top of the vocal's upper formants rather than underneath them. That is a fight for the same 2 to 4 kHz window, and static EQ handles it badly because the flute is not playing continuously. Use a dynamic EQ or a sidechained band on the sopilka, keyed off the vocal, pulling 2 to 3 dB around 3 kHz only while the vocal is present. The flute keeps its full tone in the instrumental passages, and it steps back for the lines.
Two more moves that are almost always right:
- High-pass the vocal at 80 to 100 Hz, 12 dB per octave. There is no musical information down there in an open-throat take, only stand rumble and proximity build-up.
- Leave a dip, not a boost, around 250 to 350 Hz if the take was recorded close. That is where the chest resonance turns into mud once you stack a doubled line.
Do not add a presence shelf. The singer already built one.
How do you keep it intact through a broadcast chain?
Anything heading for television or a contest stage gets measured, not eyeballed. The loudness algorithm is standardised as ITU-R BS.1770, and European broadcast delivery targets -23 LUFS integrated under EBU R128. Streaming is a different number: Spotify normalises playback to roughly -14 LUFS (their own loudness documentation), Apple Music sits nearer -16.
Two practical consequences.
Master to the medium, not to a habit. A master pushed to -8 LUFS for streaming will be turned down on playback anyway, and you will have spent your dynamic range for nothing. On a broadcast chain it will be turned down further and the limiter artefacts will still be there.
Check mono. Broadcast fold-down is still real, phone speakers are still mono, and a doubled or widened vocal that sounds enormous in stereo can lose 3 to 6 dB of level in the centre when the sides cancel. If the lead vocal is genuinely mono and the widening lives only in the doubles, this problem disappears.
Working on something with traditional vocals in it?
If you have a folk-pop hybrid that is not sitting right, the fastest way to find out why is to send the raw stems, not the mix. Get in touch and say what the record has to do: streaming, broadcast, a live stage, or all three. The chain changes depending on the answer.
Full credit list: antonchilibi.com/credits. Background: about.