Make a quiet recording properly loud

How to fix a quiet or uneven recording

1. Open the editor and drop your file in. If it is very quiet, the waveform display magnifies automatically so you can still see it.
2. Press equalize volume. Around -18 dB suits spoken word; lower keeps more dynamics, higher survives a car stereo.
3. Listen. Loud and quiet passages now sit at the same level, and the gain never moves while someone is speaking, so there is no pumping or breathing.
4. Optionally press normalize as a final step, then export.

Why this works better than a volume slider

• One gain per passage. Each spoken stretch gets a single steady gain, and changes happen only in the silences between phrases, where nothing can be heard moving.
• Room tone stays down. Silence between sentences is never amplified on its own, so the noise floor does not surge in the gaps.
• The target is absolute. Run it twice, or change your mind and pick a new target: you land exactly there, nothing stacks.
• Pair with noise cleanup: clean first, then level, and a whispery phone memo comes out clear and even.

Three different fixes people all call "make it louder"

They do genuinely different things, and picking the wrong one is why recordings often come out no better.

Tool	What it does	Right when
Normalize	Multiplies the whole file by one number until the loudest moment reaches the ceiling	The recording is already even, just quiet overall
Level	Brings each spoken passage to the same loudness independently	The volume wanders, which is the usual real problem
Compress	Continuously squashes loud moments as they happen	Music and broadcast; on speech it often pumps

Normalizing a recording that drifts achieves almost nothing: one loud moment near the start sets the gain for everything, and the quiet stretch in the middle stays quiet. That is the case levelling exists for.

What target to choose

Around -18 dB suits most spoken word and is the default. Lower, toward -24, keeps more light and shade for something listened to attentively in a quiet room. Higher, toward -14, survives a car or a phone speaker at the cost of sounding flatter. If you are delivering to a podcast host, they will normalize to their own standard anyway, so anything sensible here is fine.

The target is absolute, which matters more than it sounds. Run it twice at the same number and the second run does nothing. Change your mind and run it at a different number and you land exactly there, rather than on a compound of the two. You can experiment without digging a hole.

Why it does not pump

A naive compressor moves the gain continuously, including while someone is speaking, and the ear is very good at hearing that: it is the breathing, surging quality that makes amateur audio sound amateur. TidyAudio instead finds each spoken passage, gives that whole passage one steady gain, and confines every gain change to the gaps between phrases where there is nothing to hear moving. Silence between sentences is never lifted on its own account, so the noise floor does not surge in the pauses.

Why your recording came out quiet

Worth knowing, because prevention beats repair:

• Distance. Sound falls away sharply with it. Doubling your distance from the microphone costs roughly 6 dB, which is a lot.
• Automatic gain control. Phones and laptops constantly adjust their own input level, which is exactly what produces a recording that wanders.
• The wrong microphone. Laptops often default to a built-in array when you assumed the headset was in use.
• Over-cautious levels. Recording quietly to avoid clipping is sound instinct, but overdone it leaves the signal buried in the noise floor.

If the recording is both quiet and hissy, which is the usual pairing, clean the noise first and level afterwards.

Common questions

What is the difference between levelling and normalizing?

Normalizing multiplies the whole file by one number so the loudest moment hits a ceiling; quiet parts stay quiet. Levelling brings each spoken passage to the same loudness, which is what actually fixes an uneven recording.

Will it amplify the background noise too?

Noise under speech rises with the speech, which is why cleaning noise first is recommended. But silence between sentences is deliberately never lifted on its own.

Why can I not see my waveform at all?

Very quiet recordings used to draw as a flat line in most editors. TidyAudio magnifies the display automatically and tells you by how much, so even a peak of a few percent is visible.

Will it pump like a compressor?

No. Gain changes are confined to the gaps between phrases. While someone is speaking the gain is steady, which is exactly what a naive compressor cannot promise.

Next door