AI Voice Clone De-Noiser
Smooth out metallic artifacts and robotic sibilance from ElevenLabs and other voice cloning outputs. Multi-stage audio processing in your browser.
Frequently asked questions
How does the voice de-noiser clean up audio?+
It runs your file through a multi-stage processing chain built on the Web Audio API: a high-pass filter to cut rumble, a noise gate for background hiss, a low-pass filter for harsh top end, a de-esser targeting sibilance around 7 kHz, warmth and presence EQ, and gentle compression. The result is rendered offline and exported as a new file.
Is my audio uploaded to a server?+
No. Decoding, processing, and export all happen locally in your browser — the audio never leaves your device. That makes it safe for unreleased voiceovers and client recordings.
Will it completely remove metallic AI-clone artifacts?+
It reduces and smooths them, but EQ and dynamics processing cannot reconstruct audio detail that the voice model never generated. Heavily artifacted clips may still sound synthetic after treatment — in those cases, regenerating the audio with different voice settings usually beats post-processing.
Can I adjust the processing?+
Yes. The high-pass and low-pass cutoff frequencies, noise gate threshold, compression ratio, de-esser intensity, and breathing reduction are all adjustable, and sensible defaults are provided if you just want one-click cleanup.
What format is the output?+
The processed audio downloads as a 16-bit PCM WAV file at the original sample rate. WAV is uncompressed, so it may be larger than your input file but loses no quality.
Does it only work on AI voice clones?+
No. It is tuned for artifacts common in ElevenLabs-style voice clones, but the same chain works on any speech recording — podcast audio, voice memos, or video narration that needs hiss removal and sibilance control.