Noise reduction guide — clean recordings | Itqan

Improve recording clarity by removing noise — Itqan noise reduction guide.

You finished a 34-minute podcast episode in your bedroom — and the window AC hissed through every take. Or you saved a source interview at a busy café where espresso machines compete with the speaker. Or you filmed a tutorial outdoors and wind slaps the mic whenever you move. Itqan's audio noise reduction cleans recordings with five noise profiles and three adjustable intensity levels — up to 150 MB per file, uploaded from your device, Google Drive, or Dropbox. This guide explains how to pick the right type and strength, and how to chain the tool into real workflows before transcription or publish.

What is Itqan noise reduction?

Server-side processing that reduces background noise — AC hiss, 50/60 Hz hum, wind on the mic, or crowd ambience — while keeping speech natural. Upload, choose type and intensity, preview, download.

Unlike the audio converter (format only) or audio editor (trim/effects, no denoise), this tool improves audio quality. The audio analyzer diagnoses without editing — noise reduction is the cleanup step before editing, transcription, or publishing.

When do you need noise reduction?

  • Podcast: Home recording with AC, fan, or fridge hum in the background.
  • Interviews: Street or café noise competing with the subject.
  • Lectures: Room AC hiss or old projector buzz under the speaker.
  • Video: Polish the soundtrack before YouTube or a course platform upload.
  • Before transcription: Denoise, then run audio to text for better Whisper accuracy.
  • Music and archives: Tame hiss on old cassette or DAT transfers.
  • Zoom and call recordings: Tame keyboard clatter or PC fan hum under meeting audio.
  • After extraction: Audio from the video audio extractor often needs a cleanup pass.

Goal: make speech intelligible, not studio-perfect from a ruined take.

Noise types — selection guide

Itqan offers five profiles. Pick the closest match to what you hear — not always "general."

TypeWhat it sounds likeCommon sourceWhen to choose it
GeneralMixed, undefined background noiseTypical room, multiple sourcesYou cannot pin down the type, or several overlap
HissSteady high-frequency "shhh"Old AC, cheap preamp, bad cableNoise is constant and bright across the clip
HumLow 50/60 Hz buzzMains power, fluorescent lights, power brickYou hear a steady tone under speech — common in lecture halls
WindGusts and buffeting on the micOutdoor recording, mic without windscreenRecording on the street or in a car with air movement
CrowdAudience or public-space chatterConference, busy café, marketMultiple human voices in the background — not electronic hiss

When unsure, start with general at medium. If hum remains, reprocess with hum. Do not stack types in one session — process, listen, re-upload with a different profile if needed.

Intensity levels — low, medium, high

Intensity controls how aggressively noise is removed. Default is medium — a sensible starting point for most clips.

LevelEffectBest for
LowLight cleanup — preserves natural toneNearly clean studio takes, mild background only
MediumBalance between removal and voice preservationHome podcast, lectures, most interviews
HighAggressive removal — risk of robotic or chopped speechSevere noise; try only if medium falls short

Practical rule: start medium. If hiss remains, try a more specific type before jumping to high. Very high settings can swallow word endings.

How to denoise on Itqan — step by step

  1. Open noise reduction in a modern browser — Chrome, Firefox, or Safari.
  2. Upload from your computer, Google Drive, or Dropbox, or drag a file onto the drop zone.
  3. Wait for file name, size, and format to appear in the sidebar.
  4. Preview the audio — identify the dominant noise (hiss? hum? crowd?).
  5. Select the noise type from the dropdown.
  6. Set removal intensity — start at medium unless noise is very mild.
  7. Click Start processing and watch the progress bar.
  8. Listen on the result page; if unhappy, adjust type or intensity and process again.
  9. Download the cleaned file and keep the original on your device.

Process twice with different settings if needed — save the better take and A/B on headphones.

Podcast — home recording with AC

A host records a 34-minute episode with an aging window unit. MP3, 28 MB. Hiss sits under the voice but speech is clear.

Settings: Type hiss, intensity medium. If a faint hum remains, run a second pass with type hum at low intensity on the cleaned file.

After cleanup: Trim dead air in the audio editor, convert for distribution via the audio converter if needed.

Interview — café ambience

A reporter records a 22-minute interview in a crowded café. Customer chatter in the background. Phone saves M4A, 18 MB.

Settings: Type crowd, intensity medium. Do not jump straight to high — it may gate the source's pauses.

Pipeline: After denoising, run audio to text (Small model). Verify quotes against the original.

Lecture — AC hum in the hall

A student captures a 55-minute lecture with the laptop mic. 50 Hz hum from the HVAC runs the full length. WAV, 92 MB.

Settings: Type hum, intensity medium. Hum is steady and tonal — a better match than general here.

Tip: If the file exceeds 150 MB, compress first with the audio converter to MP3 at 128–192 kbps.

Video — polish audio before publish

A creator has a 12-minute MP4 tutorial — voiceover with light wind from an open window. 45 MB.

Settings: Upload MP4 directly (the tool processes the audio track). Type wind, intensity low to medium — wind is intermittent, not constant.

After cleanup: Replace the audio track in your video editor, or extract audio separately with the video audio extractor for finer control.

Zoom meeting — keyboard and fan hum

An employee exports a 48-minute Teams recording. The laptop fan hums under the manager's voice; keyboard clicks appear when someone types. M4A, 34 MB.

Settings: Type hiss, intensity low to medium. Fan hiss is steady; clicks may remain — acceptable for meeting minutes. Avoid high — pauses can sound gated.

After cleanup: Run audio to text (Small model), proofread names, trim dead air in the audio editor first if needed.

Before speech-to-text — why cleanup matters

Whisper — the engine behind Itqan's audio to text — tolerates noise, but pre-cleaning cuts errors on names and numbers. Steady noise (hiss, hum) confuses the model more than occasional bursts.

  • Clean first, transcribe second — not the reverse.
  • Long lectures with hum plus hiss may need two passes with different types.
  • After transcription, edit names manually — denoising does not replace proofreading.

Tips from practice

  • Use headphones — laptop speakers hide mild hiss.
  • Keep the original — every pass outputs a new file.
  • Specific type beats higher intensity — hiss at medium often beats general at high.
  • Record cleaner when you can — closer mic, closed windows, fan off before recording.
  • Chain tools: denoise → editortranscribeconvert.
  • Optional: audio analyzer shows noise in the spectrum before processing.
  • Mobile works: Upload voice memos directly — preview is clearer on phone headphones.

Pre-upload checklist

  • Confirm the file plays start to finish — corrupt uploads sometimes fail without an early warning.
  • Verify size is under 150 MB; compress or trim if needed.
  • Format is supported: MP3, WAV, AAC, FLAC, OGG, M4A, MP4.
  • Listen to 30 seconds and identify noise type before choosing.
  • Start at medium intensity — raise only if noise stays obvious.
  • For confidential or legal recordings, confirm your organization allows cloud processing.
  • Keep a copy of the original before any processing.

What the tool cannot do

  • Cannot fix completely destroyed or clipped source audio.
  • Very high intensity = robotic or chopped speech.
  • Does not separate speakers or remove intentional music.
  • 150 MB max — split or compress long WAV files.
  • Not iZotope RX — built for fast, free cleanup.
  • Each session outputs one file — to test multiple settings, process, download, then re-upload the best take.

FAQ

Is noise reduction free?

Yes — in the browser with no account or subscription. Upload, process, download.

Which formats are supported?

MP3, WAV, AAC, FLAC, OGG, M4A, and MP4 (audio track processed). For unsupported formats, convert first with the audio converter.

Are my files safe?

We process your file for noise removal only and delete it automatically after processing — not shared with third parties. See Privacy Policy and Security.

Does it work on mobile?

Yes — responsive on Android and iPhone. Upload voice memos straight from your phone.

What is the size limit?

150 MB per upload. For longer files, compress or split with the editor.

General or hiss — which should I pick?

If noise is a steady "shhh" sound, choose hiss. When unsure, start with general at medium and adjust.

Process before or after transcription?

Before — cleaning the recording improves audio to text accuracy.

Wrap-up

Upload your recording to Itqan noise reduction, pick type and intensity, download a cleaner take in minutes. Start at medium, use a specific profile when you know the source, keep the original. From bedroom podcast to lecture-hall hum, a quick pass saves re-recording — then chain through the editor and transcription tools as needed.

Security and privacy

At Itqan Tools uploads are handled per our legal pages:

  • Encrypted HTTPS (TLS) connections.
  • Files used only to complete your job — not for model training or ad profiling.
  • Automatic deletion of temporary server copies after processing.
  • Cookie and ad preferences in our Cookie Policy.

For sensitive material, confirm your organization permits cloud processing and read our Privacy Policy first.

Back to blog