Voice Enhancer Presets: Natural, Studio, Podcast, Broadcast, and Clear Compared

VocalScroll's Voice enhancer runs on your device. When you record, VocalScroll saves the raw, unaltered microphone stream, so recording never adds delay or clicks. The enhancer works afterwards: when you play a take in Takes, and when you export it.

The Voice tab of an open take offers six presets. Each one is already tuned, so you pick by sound.

---

What happens to your voice

Two things run, in this order.

Voice cleanup (one switch, on by default). It does the work you would otherwise tune by hand, and five small controls under it set how hard each part works:

1. Noise removal: a small on-device neural network takes out background noise such as fans, hiss and traffic, and keeps the air above 8 kHz in your voice. A device too slow to run it while playing uses a lighter network (RNNoise) for playback. When you export, a fast enough device uses HD voice instead (see below). The share in the table is how much of the cleaned sound goes into the mix; the rest is your original.
2. Pauses (gate): lowers the room noise left between phrases. It opens before the first syllable, so words are not clipped.
3. Breaths: lowers inhales that are quiet and noise-like. Speech is left alone.
4. S sounds (de-esser): lowers harsh S sounds that are louder than the rest of your voice. Ordinary S sounds stay as they are.
5. Echo removal: two things. One shortens the tail a room leaves after each word. From level 2 up, a second one also lowers the echo that builds up while you are speaking. It cannot tell your voice from its own echo, so at 4 and 5 it changes held vowels a little.
6. Auto level: moves your speaking level toward a steady target, slowly, so quiet and loud passages sit closer together.

The channel strip (always on, except for Original):

1. Highpass filter: cuts low rumble.
2. Hum notch: removes 60 Hz mains hum.
3. Warmth and de-boxer EQ: body around 120 Hz, less hollowness around 320 Hz.
4. Presence EQ: brings words forward.
5. Compressor: evens out loud and quiet speech.
6. Air shelf and de-hiss filter: restores clarity and trims the top end.
7. Saturation: a touch of harmonic density.
8. Peak limiter: keeps loud peaks from clipping. An export is also levelled to -16 LUFS with peaks held under about -1 dBTP, a usual level for online video and podcasts.

Each preset changes the settings, not the order. Turn Voice cleanup off to use the channel strip alone. It is switched off automatically for Original.

---

How strong: the five levels

Under the Voice cleanup switch, in Takes and in Settings, Noise removal, Echo removal, Pauses, Breaths and S sounds each have a level from 0 to 5. The preset keeps its character; the level scales what the preset does.

A preset that does not use a part at all, such as Natural's echo reduction, stays off at level 1 and starts at about 3 dB times the multiplier from level 2. VocalScroll starts at Noise 2, Echo 3, Pauses 2, Breaths 2 and S sounds 2, a little stronger than level 1, because level 1 was too light for most rooms. Original ignores all five.

Echo removal is the one to use with care. On a recording with a lot of room in it, level 4 or 5 lowers the echo by several decibels more than level 1 does. On a clean recording it has nothing to remove, so it only changes the voice.

HD voice

When you export, a device that is fast enough uses a larger noise-removal network, DPDFNet, instead of the small one. It reduces noise and some echo more cleanly than the small network. It is not a studio restoration: it cannot bring back what the microphone never caught. The model is one file of about 10 MB, downloaded from VocalScroll's own storage the first time an export needs it and kept on your device after that. On mobile data it asks first. If it is not there, too slow, or fails, the export uses the small network and carries on. The Noise removal level and the preset set how much of it goes into the mix. There is nothing to pick: the app decides.

---

Presets breakdown

1. Natural

• Sound: Your voice as it is. Part of the original sound is kept in the noise removal, the gate and echo reduction are light, and the tone is almost flat.
• Best for: Clean recordings in a quiet room, and anyone who finds stronger processing too much.

2. Studio (the default)

• Sound: Bright and polished. A lean low end, a lot of presence and open air on top, with less hiss, echo and boxiness.
• Best for: Bedroom podcasters and YouTubers recording in untreated rooms.

3. Podcast

• Sound: Warm, close and dense. A rounded low end, a hint of sparkle, noticeable saturation and firm compression, for a voice that sits right in your ear.
• Best for: Spoken-word shows and interviews you want to sound even from start to finish.

4. Broadcast

• Sound: Announcer. A big chest, a mid-forward voice, a softer top and very even dynamics from heavy compression.
• Best for: Voiceover, announcements and audio-only shows.

5. Clear

• Sound: Thin and forward on purpose. The low end is cut hard, presence is high and the top is trimmed, with the deepest gate and the most echo reduction.
• Best for: Noisy or echoey rooms, and laptop, webcam and phone microphones. Speech cuts through.

6. Original

• Sound: The recording exactly as captured. No cleanup, no processing, no levelling.
• Best for: Hearing your hardware's raw output, or keeping a file untouched.

Settings by preset (level 1)

Pick a preset on the Voice tab, then open the Export tab and choose Export video or Export audio (a WAV) with Mastered voice on to save the result.