Clean a noisy phone recording: how to make a conversation easier to hear
Identify noise in phone and voice-recorder conversations, compare profile-based and speech-focused treatment, and review the result.
A noisy conversation, an intrusive click, a room tail between words. Open your own audio in the browser, choose a tool and hear what the treatment changes.
Free local editing and Preview. Paid WAV export, priced for your work.

The waveform reveals changes over time; the spectrogram shows frequency content. Zoom into a problem, select the passage to treat and audition settings with a short Preview.
Input, Result and Residual refer to the same passage. Check whether processing removed useful speech as well as interference. Applied steps remain separately visible and reversible.
Understand the controlsStart with what you hear. These guides help you choose a tool, try a treatment and recognise when to ease off.
Identify noise in phone and voice-recorder conversations, compare profile-based and speech-focused treatment, and review the result.
When to use De-click or spectral attenuation, and how to preserve speech.
Distinguish room decay from repeated echoes, then set up and review De-reverb.
Separate level, noise and voice balance: when gain is enough and when another tool is needed.
Identify clipped peaks, evaluate estimated repair and leave output headroom.
Two estimated speech sources or reference-guided extraction: which to try and how to review it.
From denoising to local spectral editing. Each manual explains the task, parameters and listening checks.
Opening, analysing and processing a local file happens in your browser. First use may download processing assets. Larger recordings need free storage and memory; on a phone, start with a short passage.
Editing and Preview are free. Before export, the itemised calculation shows the price for the current work; purchase requires an account and billing details. Sending audio to a specialist is a separate explicit upload.
The editor offers repair methods; results depend on the source. It cannot guarantee recovery of information lost beneath noise or authenticate an uncertain word.