A bandura and a string section want the same band, roughly 400 Hz to 3 kHz, so this mix gets decided before any equaliser is opened: what the arrangement leaves empty, where the two microphones stand, and whether the attack of a plucked metal string is allowed to arrive ahead of the bowed sound behind it. I mixed and mastered the orchestral version of Maryna Krut's "Скажи мені", and most of the useful work on a record like that is subtraction. A Kyiv-style bandura carries 55 to 65 metal strings tuned chromatically across five octaves. Put an orchestra behind it and most of those strings stop existing.
What is actually on the instrument?
More strings than any other plucked instrument you are likely to record, on a soundboard the size of a small guitar.
The standard concert instrument, built for the Kyiv playing style, has 12 to 17 bass strings running along the neck and around 50 treble strings, the prystrunky, stretched in two levels across the soundboard like the black and white keys of a piano. The Ukrainian Bandurist Chorus lists a prima at 12 basses and 43 trebles covering almost five octaves, and concert instruments at 62 to 65 strings. All of them are metal. Gut went out after 1891 and steel was standard by 1902, which is why the decay is bright and fast rather than the soft one people expect from a folk instrument.
Two consequences for recording. Sound radiates from a wide surface rather than a hole, so microphone position changes the balance between basses and prystrunky far more than it changes tone. And players use artificial fingernails, in common use since the mid-1970s, so a nail on steel gives a real transient with energy past 8 kHz. That transient is why the instrument can be heard against an orchestra at all.
Where does the bandura collide with the orchestra?
In the middle, and in the attack.
Violins and violas occupy the same 400 Hz to 3 kHz region where the prystrunky work, and they fill it with continuous bowed energy rather than decaying plucks. Two seconds into a sustained string chord, a plucked note that started 4 dB above the section is 12 dB below it, because the bandura note decays and the section does not. Static level balance is therefore lying to you: the instrument is present on a chord change and gone mid-phrase, and the meters show nothing.
The second collision is the low end. Bass strings, cellos and double basses share an octave, and unlike the mid-range problem this one is audible as mud rather than as absence. A high-pass on the bandura at 90 to 120 Hz costs almost nothing musically and buys the low register back.
Why does equalisation fail here, and what works instead?
Because the frequencies you would cut are the ones carrying the melody.
The instinct is to notch the string section where the bandura sits. Do that on a real orchestral recording and the section goes thin and papery, because a violin section without 1 to 3 kHz stops sounding like violins. The masking is not a resonance. It is two sources genuinely playing the same notes in the same octave.
Three things work, in this order.
Arrangement first. Ask the arranger for gaps rather than asking the mix for them. A string section that rests for the two beats where the bandura figure lands solves the problem completely and costs nothing. Once the arrangement is committed the option is gone, which is why it gets asked before the orchestra is recorded.
Dynamic equalisation keyed off the bandura. A 3 to 4 dB dip on the string bus centred around 1.5 kHz, with a wide Q, triggered by the bandura bus rather than by the string bus itself. It opens only when the instrument plays and closes in the rests. Attack fast, around 5 ms, release slow, 250 to 400 ms, so the section does not pump on every pluck.
Transient shaping on the bandura rather than level. Adding 3 to 5 dB of attack while leaving sustain alone puts the instrument in front of the orchestra without making it louder. It is the most effective single move here, and the one most engineers skip, because the reflex is the fader.
Where do the microphones go?
Two, at different distances, and neither of them close.
A small-diaphragm condenser about 40 cm above and behind the player's right shoulder, aimed across the prystrunky rather than at a point on them, gets the treble register with the nail attack intact. A second microphone 60 to 80 cm out in front, level with the lower soundboard, gets the basses and the body. Balance those two and you have the instrument.
Closer than about 30 cm and you record whichever strings happen to be nearest the capsule, and the balance changes every time the player's hand moves. That is the most common mistake with this instrument, and it is unfixable in the mix because there is no single correct level for the file.
Check polarity between the two microphones on a plucked bass note rather than on speech, in mono. If the distance difference is more than about 40 cm, delay the near one by roughly 3 ms per metre before deciding anything about tone.
What happens when the singer is also the player?
The instrument is already in the vocal microphone, so it stops being a separate track.
Bandurists sing while they play, which means the vocal microphone contains a bandura at maybe 8 to 12 dB below the voice, arriving a millisecond or two early or late depending on the geometry. Gating it does nothing useful. Editing the vocal moves the bandura underneath it and produces jumps in the bleed that are audible as a small click at every edit point.
Treat vocal and instrument as one performance from the start. Comp full passes rather than syllables. Use a figure-of-eight vocal microphone turned so one of its side nulls, at 90 degrees off axis, faces the soundboard, worth 10 to 15 dB of rejection for no money. And after processing the instrument tracks, listen to the vocal alone, because heavy transient work on the bandura bus will not match the bandura sitting inside the vocal track and the two will fight.
What changes at mastering?
The plucks reach the limiter first, and they are not the loudest thing musically.
A record with a plucked steel instrument up front has short peaks 6 to 8 dB above the orchestral body. A limiter set for the orchestra catches the nails and dulls them, and the moment the attacks go dull the instrument slides back behind the strings and everything gained in the mix is gone. Two ways out: a slower limiter release, 200 ms and up, so gain reduction does not follow individual plucks, or a light clip on the bandura bus in the mix so the peaks arrive already tamed by a stage that was chosen rather than accidental.
Loudness targets deserve less respect here than on pop. An orchestral arrangement pushed to a fixed integrated target gets compressed until it stops being orchestral, and no platform rewards that.
Frequently asked
Can you just use a pickup and skip the microphones? A pickup solves the bleed and creates a tone problem, because the soundboard is where most of the instrument happens. Record both: pickup for transient recovery under the orchestra, microphones for everything else.
Does the bandura need reverb if the orchestra was recorded in a hall? Usually less than you expect, and never the same reverb. Send it to the orchestral space at a low level so it shares the room, then keep the direct sound dry so the attacks stay readable.
Should the bandura be centred? Centre it if the player is also the singer, because the voice is already there and the bleed will pull the image anyway. If they are separate performers, 15 to 20 degrees off centre keeps both readable.
If you have a folk instrument fighting an arrangement
Send me the session and say which instrument disappears and where. Usually the answer is two arrangement decisions and one about attack, not a chain of plugins. Full credit list at /credits, background and contact at /about. Maryna Krut tours the instrument internationally if you want to hear what it does live.