Record a line or upload a clip, and the ghost voice changer turns it into a breathy, trembling whisper drifting down a long, empty hall. Compare it with your original and download a WAV. Free, private and right in your browser.
How it works
Three steps. You don't need to whisper; the effect does that part for you.
Step 1
Press Record and speak softly for up to 60 seconds, or upload an MP3, WAV, M4A or OGG file up to 20 MB and 5 minutes long. A quiet room matters here, because the reverb makes any background noise linger.
Step 2
Ghost is preselected, so press Apply effect. Your browser turns your voice into breath and builds the haunted hall around it in a few seconds, all on your own device.
Step 3
Flip between Original and Ghost while it plays, then download ghost-voice.wav. The echo and reverb tail is included, so your last word drifts away instead of stopping.
Scripts to try
Slow, quiet sentences with long pauses are the creepiest. Leave room for the hall to answer.
Who's there? I've been waiting for you... for so long.
Get out... get out of my house...
Loop it on a hidden speaker with a few seconds of silence between repeats.
It happened on this very night, one hundred years ago.
Take one piece of candy... only one... I'm watching.
Is anyone here with us? Knock once for yes.
Highlights
A whisper engine, a tremble and an empty hall, stacked in a fixed recipe.
The effect keeps the shape of your words, the way your mouth filters sound, but swaps the buzz of your vocal folds for breath noise. Eighty percent whisper with a trace of your real voice underneath sounds like someone who is barely there.
Pitch wavers about 4.5 times a second while the volume pulses 5 times a second. Because the two rates differ, the wobble keeps drifting in and out of step and never quite repeats.
A low cut at 260 Hz strips the body from the voice. Then a 230 ms echo and a 3.2-second reverb, mixed almost as loud as the voice itself, put it at the far end of a long, empty corridor.
Quick reference
The exact chain behind the ghost voice. Every take gets the same treatment, so your lines match.
| Stage | Setting | What you hear |
|---|---|---|
| Whisper | 80% breath, 20% voice | Airy, voiceless speech |
| Vibrato | 4.5 Hz wobble | Wavering pitch |
| Tremolo | 5 Hz, 25% depth | Pulsing, fading level |
| Filters | Low cut 260 Hz, softer highs | Thin, distant, bodiless |
| Echo | 230 ms, 30% feedback | Repeats down the hall |
| Reverb | 3.2 s, 50 ms pre-delay | A big, empty space |
Your pitch and timing are not changed, and the settings are fixed. There is no reverse-reverb option; for a reversed swell, reverse the downloaded file in any audio editor.
A whisper is speech without vocal fold vibration: air rushes through a partly open throat, and your mouth still shapes it into words. That is why you can understand a whisper, yet it hides most of what normally tells you who is talking, such as pitch, age, warmth and much of the emotion. A voice you can follow but cannot place is unsettling on its own.
The rest is space and instability. A long reverb with no clear source tells your brain the speaker is somewhere large, empty and far away, and a constant tremble suggests something that cannot hold still. Horror sound design leans on the same cues: a voice you can hear but cannot locate.
Under the hood, the whisper uses linear prediction (LPC), a technique from early speech synthesis and phone codecs. Every 25 milliseconds it estimates the filter your vocal tract applied, then drives that filter with noise instead of your voice. The words survive; the voice behind them does not.
Speak softly, not in a whisper. The effect needs a clear voice to analyze. Real whispering gives it less to work with and brings up room noise, so use a soft, slow speaking voice and let the effect turn it into breath. Stretch the S and H sounds and leave long pauses for the echoes.
Keep the room quiet. Fans, traffic and fridge hum get the same long reverb as your voice and turn into a ghostly roar. Record about 6 to 8 inches (15–20 cm) from the mic, a little off to the side so P and B sounds don't pop.
Set up the scare. For a porch or yard, loop the WAV on a Bluetooth speaker hidden in a bush or behind a door, with a few seconds of silence between lines. For videos and podcasts, layer it quietly under the scene. For a possessed voice, stack it with the same line from the monster voice changer.
Rather type? The scary voice generator creates creepy narrators and demon voices from text with AI; it needs a free account and uses credits. The ghost voice changer is free and haunts your own recording. Keep scares for people who are in on the fun, and never use a disguised voice to frighten or threaten someone for real.
Last updated October 21, 2026
Use cases
Wherever you want people to lean in closer, and then regret it.
Porch greetings, graveyard props and haunted-house loops on a hidden speaker.
Spirit voices, cold opens and campfire-style stories for audio shows.
Whispering apparitions, found-footage voices and jump-scare setups.
Eerie clues and voices from inside the walls for themed puzzle rooms.
FAQ
Straight answers about the haunted sound, Halloween setups and privacy.
Still have a question?
The quickest way to find out is to try it yourself. It only takes a minute.
Record a line, apply the ghost effect and download it in about a minute. Free, no sign-up, nothing uploaded.
