Remove background noise from your podcast, on your phone

Record one voice per track when you can

Noise Remover keeps what sounds like speech and turns down what does not. A track that carries a single voice gives it the cleanest job, so the way you record a show matters as much as the cleaning.

Solo episodes

Record in the app with the phone about 15 cm (6 in) from your mouth and slightly to the side, so your breath doesn’t hit the microphone. Turn off what you can (the fan, a humming fridge) and leave the rest to the cleaning. The recorder captures 48 kHz mono, with pause, a timer and a level meter.

Remote guests

Ask each guest to record their own voice on their own phone while you talk over a call on another device (a computer or a second phone), since a phone busy with a call usually cannot record at the same time. Wear headphones on both sides so neither voice leaks into the other microphone, and start with a clap or a countdown. Each file then holds one voice: clean them one by one, and line them up in your editor on the clap.

Guests in the same room

With one phone in front of each speaker, every track still hears the other voice faintly. That is fine while you take turns. What suffers is crosstalk: when you talk over each other, the cleaning keeps both voices, and can turn the main one down a little. Let each other finish: a short pause before an answer costs nothing in the edit.

How do I clean podcast audio on my phone?

1. Bring in each track

   Import each speaker’s file, or share it to Noise Remover from your recorder or your messaging app. Cleaning starts as soon as the file arrives.

2. Listen at the same spot

   Switch between Original and Cleaned while it plays: the switch keeps the position, so you hear exactly what changed. Check a quiet passage and the ends of sentences, where over-processing shows first.

3. Set the intensity

   The default 80% caps the reduction near 29 dB. Lower it if your voice sounds thin; take it to 100% for a very noisy room.

4. Keep the level on

   Ready-to-publish level, the app’s loudness normalization, brings the track to −16 LUFS with a −1 dBTP ceiling, so voices and episodes land at a consistent loudness.

5. Trim and export

   Cut the count-in and the “are we recording?”, then export an M4A (48 kHz, 128 kb/s, no watermark) and share it to your editor or save it to your phone.

If you mix the episode in an editor with music and several voices, export WAV instead: the editor works on uncompressed audio, and the sound is encoded only once, when the finished episode is exported.

With Pro, batch cleaning takes every track of an episode in one pass, with the same settings, instead of one file at a time.

Set your usual intensity, level and low-cut once in Settings, and every new track starts from them.

Reopening a track from History to change its intensity, level or trim, and exporting it again, never counts as a new cleaning, which is useful when the edit moves on after the first export.

Level every track to −16 LUFS

Listeners move from one show to the next without touching the volume, so an episode published too quiet sounds weak beside the others. With Noise Remover, Ready-to-publish level measures the cleaned voice and applies one gain to bring it to −16 LUFS, a target widely used for spoken-word podcasts (see LUFS and normalization explained).

When each speaker’s track is leveled on its own, the voices come out balanced against each other, so you spend less time riding faders through the conversation.

There is no limiter. If a track has sharp peaks (a laugh, a hand landing on the desk), the gain stops at the −1 dBTP ceiling and the track lands a little under −16 LUFS: nothing is squashed to get there. Once you add music or mix several tracks in an editor, check the loudness of the finished episode there.

Keep your voice sounding like you

Over-processing is what makes a cleaned voice sound robotic or underwater. The intensity slider sets how far Noise Remover may turn the noise down: at the default 80% the reduction stops near 29 dB to keep the voice natural, and at 100% it is uncapped. Your ears decide, and the Original/Cleaned switch is how they compare. More in why noise reduction can sound robotic.

If the cleaned voice comes out more than 6 dB quieter than the original (usually because another voice overlapped it), the app shows a tip under the slider: a lower intensity keeps it natural.

Long episodes and the free plan

On the free plan, Noise Remover cleans the first 3 minutes of each file, 3 files a day, with ads: enough for a trailer, an intro or a short segment, with no watermark and the same quality as Pro. A full episode runs longer:

• One longer file: watch an optional video on the result screen to clean that file up to 15 minutes.
• Every episode, at any length: Pro removes the length limit, the daily limit and the ads. Plans and prices are shown in the app (see Free and Pro).

What cleaning can’t fix in a podcast

• Crosstalk. In our lab test at full intensity, a second voice talking over the speaker was reduced by only 0.4 to 1.6 dB, so it stays in the cleaned track (how we measured).
• Music under the voice. It is pushed down and altered. Clean the dry voice first, then add the music in your editor.
• Room echo. It is not removed, so a bare room still sounds bare. Soft furnishings and a shorter distance to the microphone do more than any setting. See echo or background noise?