How to even out quiet and loud parts in your recording
- Step 1Open the tool — Go to the Speech Leveler. Drop your recording onto it. It accepts the common audio types: MP3, WAV, FLAC, M4A, OGG, and Opus. If your recording is inside a video, see the video note below.
- Step 2Leave the preset on Podcast — There is one setting: a Preset menu. For most uneven recordings the default Podcast preset is the right choice — it evens things out gently and naturally. You do not need to change anything else.
- Step 3Try Voiceover if it is still too jumpy — If after leveling the volume still swings a lot, run it again from the original (not the leveled copy) with the Voiceover preset, which evens out more strongly.
- Step 4Use Interview for two people — If your recording has two voices — one close to the mic, one further away — pick Interview. It is tuned to bring two different voices to a similar volume.
- Step 5Click process and download — Press the process button, wait a few seconds, and download the evened-out file. Play it back start to finish to check the quiet parts are now audible.
- Step 6Listen on a phone, not just headphones — Headphones hide uneven volume. Play the result on a phone speaker — if every line is now clear there, you are done. If you also need it at a set loudness for a platform, run it through the loudness-normalizer.
Which preset should I pick?
Plain-language guide to the three presets. There are no other settings to touch.
| Your recording | Pick this preset | What it does |
|---|---|---|
| One person talking, volume wanders a bit | Podcast (default) | Gently makes quiet parts louder and loud parts quieter — the natural, safe choice |
| One person, volume jumps a LOT (whisper to shout) | Voiceover | Evens out more strongly so even big swings come together |
| Two voices, one near the mic and one far | Interview | Tuned to pull two different-distance voices to a similar level without sounding squashed |
This tool vs. what you might actually want
Evening out the ups and downs is compression. Other common requests need a different tool.
| What you want | Use this instead | Note |
|---|---|---|
| Just make the whole thing louder | volume-normalizer | If volume is even but too quiet overall, you want normalisation, not leveling |
| Set it to the right loudness for YouTube/Spotify | loudness-normalizer | This tool evens the ups and downs but does not hit a platform target |
| Get rid of background hum/hiss | ai-noise-reducer | Do this BEFORE leveling, or the leveler will make the noise louder too |
| Cut the long silent gaps | silence-stripper | Removes dead air; leveling does not shorten anything |
Cookbook
Real-world recordings and the simplest preset that fixes them. No numbers to enter — just which preset and what to expect.
A lecture you recorded from the back of the room
The professor's voice rises and falls; questions from the audience are nearly inaudible. The Podcast preset lifts the quiet bits up.
File: lecture.m4a
Preset: Podcast
Result: quiet audience questions become audible;
loud emphatic moments no longer blast
Output: lecture-leveled.m4a (same format as the input)A voice memo that whispers then shouts
You were thinking out loud, sometimes mumbling, sometimes excited. Big swing — Voiceover evens it out harder.
File: memo.mp3 First try: Podcast -> better, but still some jumps Second try (from original): Voiceover -> even throughout Tip: always re-run from the ORIGINAL, not the leveled copy
A two-person chat at a kitchen table
One phone, two people at different distances. Interview is built for exactly this.
File: chat.wav
Preset: Interview
Result: the far person is brought up toward the near person;
neither voice sounds crushed
Output: chat-leveled.wavBackground fan got louder after leveling
Common surprise: making quiet parts louder also makes quiet background noise louder. Fix the noise first.
Wrong order: level -> noise is now loud and obvious
Right order: ai-noise-reducer (remove fan/hum)
-> then Speech Leveler (Podcast)
Now the quiet parts are clean before they get lifted.Audio is inside a video file
Speech Leveler works on audio files, not video. Pull the audio out first.
Have: clip.mp4 (uneven voice) Step 1: video-to-wav -> clip.wav Step 2: Speech Leveler (Podcast) -> clip-leveled.wav Step 3: re-add the audio to your video in your editor
Edge cases and what actually happens
Needs a Pro plan to run
Pro requiredSpeech Leveler is a Pro feature, so on a Free account it stops with 'Speech Leveler requires a Pro subscription.' before doing anything. The processing is still fully in your browser with no upload either way — only the plan unlocks the tool. Free covers audio up to 50 MB / 30 minutes; Pro raises that to 200 MB / 120 minutes.
Recording is too big or too long
RejectedOn Pro the limit is 200 MB and 120 minutes per file. A very long or very large recording is turned away. You can split it first with audio-splitter, or upgrade to Pro + Media for unlimited length and up to 100 GB.
Background noise got louder
ExpectedThis is normal: making quiet parts louder also lifts quiet background sounds like a fan or hum. Remove the noise first with the ai-noise-reducer, then level. It is working as intended, not broken.
It is still too quiet overall after leveling
Use volume normaliserLeveling evens the ups and downs but does not necessarily make the whole thing louder. If it is even but still quiet, run volume-normalizer afterward to bring the overall level up.
Result sounds flat or squashed
CautionUsing Voiceover on a recording that was already pretty even can over-do it and sound lifeless. Start with Podcast; only step up to Voiceover if Podcast was not enough — and always from the original file, not a leveled copy.
You dropped in a video file
Audio onlyThis tool reads audio and gives back audio. To even out the voice in a video, extract the audio first (for example with video-to-wav), level that, then re-add it in your video editor.
The output is a different size than expected
By designLeveling re-encodes the audio, so the file size can change a little. The format stays the same as your input (MP3 in, MP3 out). The length is unchanged — leveling does not cut anything.
Two voices still uneven after Podcast preset
Try InterviewThe Podcast preset is tuned for one voice. With two people at different distances, switch to the Interview preset, which is designed to bring two different voices to a similar volume.
Frequently asked questions
Do I need any audio editing experience?
No. You drop the file, leave the preset on Podcast (or pick Voiceover/Interview), and download. There are no technical settings to understand.
Is my recording uploaded to a server?
No. Everything happens inside your browser using FFmpeg 8.1. Your voice memo, lecture, or interview never leaves your computer.
What is the difference between this and making it louder?
This evens out the difference between quiet and loud parts (compression). Making the whole thing louder is a separate job — use volume-normalizer for that. They are often used together: even it out here, then turn it up there.
Which preset is safest to start with?
Podcast, the default. It evens things out gently and naturally. Only switch to Voiceover if the volume still jumps a lot, or Interview if there are two voices.
Will this fix the loudness for YouTube?
Not by itself. This tool evens the ups and downs but does not set a platform loudness target. After leveling, run the loudness-normalizer and pick the YouTube/Spotify preset for the correct level.
What file types work?
MP3, WAV, FLAC, M4A, OGG, and Opus. If your audio is inside a video file, extract it first with video-to-wav.
Will it change my file format?
No. The output keeps the same format as your input — drop an MP3 and you get an MP3 back. There is no format setting to choose.
Why did the background noise get louder?
Because the tool makes quiet parts louder, and quiet background noise is one of those quiet parts. Remove the noise first with the ai-noise-reducer, then level.
How long does it take?
Usually under a minute for a normal recording, running on your own computer. There is no upload wait or processing queue.
Does it cut out the silent gaps?
No, leveling does not shorten anything — it only evens out the volume. To trim long silences, use silence-stripper.
Can I undo it?
The tool downloads a new file and leaves your original untouched, so just keep the original. If a result is too squashed, re-run from the original with a gentler preset.
Is it really free?
There is no upload fee, queue, or watermark — but the tool itself is a Pro feature. The Free plan can use other audio tools; Speech Leveler needs Pro. Everything still runs locally in your browser regardless of plan.
Privacy first
Every JAD Audio tool runs entirely in your browser via FFmpeg (WebAssembly) and RNNoise. Your audio files never leave your device — verified by zero outbound network requests during processing.