Support

How do I enhance a recording?

Import a file — the app works on recordings you already have and has no recorder of its own. Pick an audio file from Files (m4a, wav, mp3, aac) or, with Premium, a video (mp4, mov). Tap Enhance, and when it's done, compare the versions and export the one you want.

What's the difference between Quick Clean, Natural, and Studio?

• Quick Clean is an EQ and dynamics filter that runs on your iPhone. It's fast, it's free, it never uploads anything — and like every filter, it can only subtract what's already there. It is not AI, so we don't call it AI.
• Studio is a generative speech model running in the cloud. It doesn't just remove noise; it rebuilds the voice, which is why it can recover detail your microphone never captured.
• Natural is the same Studio run, held back. One run returns both, so you can switch instantly if Studio reaches too far.

How does the A/B player work?

Tap between Original, Quick, Natural, and Studio and playback stays at the same spot in the file. The playhead doesn't jump, so the only thing that changes when you switch is the sound itself — which is the whole point of comparing.

How long does it take?

Roughly 4 minutes for a 5-minute recording. You don't have to sit and watch it — the app sends a notification when the job finishes.

What exactly is free?

3 Studio previews, 60 seconds each — 3 in total, ever. Not 3 per day. We'd rather you read that here than discover it after the third one. They exist so you can judge the result on your own recording before spending anything, and you can export them as M4A. Quick Clean and the A/B player are free and unlimited. There are no ads anywhere in the app.

Why only three?

Each Studio run costs us real GPU time on a third-party service. Three 60-second previews is what we can give away per person while still being able to run the thing. It's enough to answer the only question that matters: does this fix your recording?

What does Premium include?

30 audio minutes per month, files up to 10 minutes, video support, lossless 44.1 kHz WAV export, and unlimited history. Each run bills a minimum of 60 seconds, and any job that fails is refunded to your pool. Minutes don't roll over.

How much is it, and is there a trial?

$7.99/month, or $39.99/year — a 58% saving. The 7-day free trial is on the annual plan only. Those are US prices; your storefront's price is shown in the app before you buy. Cancel before the trial ends and you aren't charged.

How do I restore a purchase?

Open Settings in the app and tap Restore Purchases. This re-applies any active subscription bought with your Apple ID, including a legacy Voice Isolation subscription at its original price.

How do I cancel?

iOS Settings → your name → Subscriptions → AI Voice Enhancer → Cancel Subscription. It stays active until the end of the period you've paid for. Deleting the app does not cancel a subscription.

Can I get a refund?

Refunds are Apple's to issue, not ours — request one at reportaproblem.apple.com. If a job failed inside the app, its minutes come back to your pool automatically; that's separate from a refund.

What files can I use?

Audio: m4a, wav, mp3, aac, up to 30 MB. Video: mp4, mov, up to 500 MB, with Premium. Files can be up to 10 minutes long on Premium; free previews cover the first 60 seconds.

Does it change my video?

No. We extract the audio track, enhance it, and put it back. The picture is never re-encoded — it comes out exactly as it went in.

The result sounds robotic or wrong. What now?

Switch to Natural. It's from the same run and is deliberately more conservative, and for many recordings it's the better result. This is a real limitation of generative speech models, not a defect in your file: a model rebuilding a voice will sometimes reach too far, and the more damaged the input, the more it has to invent. If neither version works for your recording, that's what the free previews are for — find out before you pay.

Can I get my original back?

You never lose it. The app doesn't overwrite the file you started with.

Does it work on music or singing?

The model is trained on speech, not music. Music, singing, and crowded multi-speaker recordings are outside what it's built for and it may degrade them. Don't run an irreplaceable recording — a wedding, a performance — through it and expect it to help.

My job failed.

Its minutes are returned to your pool automatically, so a failure doesn't cost you anything. Studio depends on a third-party GPU service that can be slow or briefly unavailable; if it keeps failing on the same file, send us the details.

Where does my recording go?

It depends on which mode you run, and the difference is real. Quick Clean stays on your iPhone. Studio does not — it uploads your audio to a third-party GPU service (Replicate, running Resemble AI's speech model) and fetches the result back over a private, unguessable link. If you never run Studio, no audio leaves your device. The Privacy Policy spells out who touches what and for how long.

Do I need an internet connection?

For Studio, yes — it runs in the cloud. Quick Clean works offline.

How long are my files kept?

Uploads are configured to expire after 1 day and results after 7 days. Copies held by Replicate while processing fall under their retention policy. To have something removed sooner, email privacy@bulpara.com with the approximate time of the job.

Do you have my account?

There are no accounts and nothing to sign up for. To meter your monthly minutes we store an identifier Apple issues for your subscription, plus how many minutes you've used. Not your name, not your email, not your Apple ID.