Mastering vocal-led acoustic Indian tracks

Loudness targets for ghazals and acoustic Indian songs

A quiet second line should still feel quieter than the phrase before it. For a ghazal or acoustic Indian song, aim for streaming loudness without erasing that contrast. Dynamic range, the difference between quieter and louder passages, is part of what keeps the voice expressive.

• Aim for integrated loudness, measured across the whole track, around minus 14 LUFS. LUFS means loudness units relative to full scale.
• Treat about minus 12 LUFS as an upper limit, not a target you must reach.
• Keep true peak, the maximum reconstructed signal level, at or below minus 1 dBTP. dBTP expresses that true-peak level in decibels.

Streaming platforms use loudness normalisation to adjust playback towards their own reference levels. These targets are practical starting points, not permanent platform rules. A heavily flattened master gains no loudness advantage when playback is turned down, while a quieter master can retain its expression.

Why heavy limiting flattens vocal ornaments

A quick shake or slide in the voice carries expression through changes in level, not just pitch. These ornaments need room to move. A limiter, a processor that controls peaks, can even out those changes when it works hard on every phrase.

Heavy limiting also reduces dynamic range, the difference between quiet and loud passages. A quiet second line can lose its contrast with the stronger line before it. Underneath, the tanpura drone, the sustained background tone, may swell and shrink with the singing instead of staying steady. The voice loses movement while its backdrop starts moving too much.

Use the limiter for peak control, with only a few decibels of gain reduction, meaning the amount it turns the signal down. Let gentle compression, a light reduction of level differences, and tonal adjustments support the level instead. The final limiter should catch peaks, not reshape every vocal phrase.

Control sibilance without dulling the instruments

Listen next to the s and sh sounds at the ends of words. Close microphones can exaggerate these sounds in Hindi, Urdu, Malayalam and Tamil vocals. Start with a de-esser, a processor that reduces harsh s and sh sounds, on the vocal in the mix.

Treating the voice there lets you control the consonants without processing the tanpura and tabla alongside them. Keep the treatment focused on the sounds that spit, rather than making the whole vocal darker. The aim is softer sibilance, the hiss of those consonants, not a duller performance.

Then listen to the finished mix before adding anything to the master. If the s sounds still jump out, use a second, gentler de-esser at this stage. Do not reach for a broad cut across the high frequencies. That can dull the tanpura's shimmer and soften the tabla's initial strike.

Keep the tanpura and tabla low end steady

Next, follow the tanpura’s lowest string beneath a vocal phrase. It shares the low frequencies with the tabla’s bayan, its bass drum. Clean up this region without taking away the weight that supports the voice.

Roll off rumble below the drone’s fundamental, the lowest frequency of its note. Let that pitch guide the cleanup rather than choosing a fixed cutoff for every song. Keep the low end mono, meaning centred rather than spread between left and right. If the master feels thin, try a gentle low shelf, a broad adjustment of the bass region, rather than simply adding a bass boost.

Compression reduces level differences, but it should not make the drone follow every rise in the singing. If the drone swells and shrinks with each phrase, the compressor is acting too fast or reducing too much. Use slower, lighter compression and listen again.

Build a gentle mastering chain

Build the chain around the voice, not the loudness ceiling. Each stage should do a little work, leaving the slides and quick ornaments expressive.

• Start with corrective equalisation (EQ), or frequency balance adjustment. Filter rumble below the drone’s fundamental, its lowest note. Use one or two narrow cuts for boxiness or a resonant note.
• Follow with compression, which controls level variation. Use slow, light settings and aim for one to two decibels of gain reduction. The drone should not swell with each phrase.
• Next, use broad tonal EQ for air and warmth. Keep the changes gentle rather than asking this stage to supply all the level.
• Finish with a limiter, which restrains peaks. Use true-peak control for the last decibel or two. True peak estimates the highest level, including between digital samples.

If adding stereo widening, which spreads sound across left and right, apply it only to reverb and instruments, never the voice or drone.

Check the master on phones and earbuds

Now move to earbuds and a phone speaker. Listen at a normal volume, including the quietest passage and a loud vocal line. Keep the voice, drone and consonants in focus as you switch between them.

• Voice: Check that the singer stays in front on both devices. Slides, shakes and quick ornaments should keep their movement rather than sound flattened.
• Drone: Listen beneath each phrase. The tanpura should stay steady, not swell and shrink as the singer gets louder.
• Consonants: Listen for s and sh sounds that stay soft. They should not spit, but the tanpura's shimmer and tabla's attack should remain audible.
• Bass: Check in mono, with left and right channels combined. Keep the low end centred, and listen for a steady balance between tanpura and tabla.
• Quiet passages: Play the quietest section without reaching for the volume control. It should remain audible while still feeling quieter than the strongest lines.

Use an online mastering tool with a clear target

Carry that contrast into the online mastering settings. Choose an acoustic or vocal setting, with a target around minus 14 LUFS. LUFS is a unit used to measure loudness; here, the target describes the average across the whole track. An online tool can handle this material well when its settings suit the voice and the space around it.

The BeatRootRaga tool lets you set the target loudness and preview the result before committing. Compare that preview with your original, unmastered mix at matched loudness, not matched peak level. The useful test is what happens to the singing and accompaniment, not which version sounds louder.

• Play a loud vocal phrase. Check that slides, shakes and quick ornaments retain their expression.
• Listen beneath the voice. The tanpura drone should stay steady rather than swell and shrink.
• Check s sounds on earbuds, then the quietest passage at your normal listening volume.

Listen for the balance in a ghazal recording

Use Gayatri Asokan’s Urdu ghazal, Kitne Aish Se Rahte Honge, as a listening reference for that balance. Listen for three things: a voice that stays in front, a steady drone, and loud lines that keep their expression.

• Voice: listen for slides, shakes and quick ornaments, the small movements that shape a vocal line. Their changes in level are part of the expression.
• Drone: listen beneath the louder phrases. It should remain steady rather than seem to breathe in response to the singer.
• Loud lines: notice whether the phrase keeps its shape instead of becoming flat. Bring that same listening test back to your own master.