How to record clean voiceover with just your phone
If your briefs keep asking for a talking-head version and a voiceover-over-b-roll cut, you're not imagining it — VO has quietly become a standard line item. Good news: the phone in your hand records VO clean enough to ship. Brands aren't buying broadcast polish — they're buying a person.
The setup:
- Use the free recorder app — Voice Memos on iPhone (Settings → Voice Memos → Audio Quality → Lossless), or your Android's built-in recorder. No paid app sounds meaningfully better.
- Airplane mode on. One notification buzz mid-sentence is a full retake.
- Hold the phone a fist-width from your mouth. Closer turns boomy and poppy; farther picks up the room.
- Speak past the mic, not into it. The mic sits on the phone's bottom edge — point it just off to the side of your chin so your breath doesn't pop the P's.
- Record full passes, top to bottom, two or three times — not one "perfect" sentence at a time. Consistent energy beats any single flubbed word.
A budget mic is a lovely later upgrade, not a prerequisite. Room and read matter more than gear.
The room setup that kills echo (no foam panels)
A closet full of hanging clothes, or a duvet draped over you and the phone, kills echo better than any foam panel you can buy — soft, heavy fabric absorbs reflections; thin foam squares mostly don't.
Quick test for any spot: clap once. A ring trailing off after the clap means the room is too live — and that ring is most of what makes home VO sound amateur.
Your three free vocal booths:
- A walk-in or stuffed closet. Fabric on three sides, phone inside with you.
- The duvet fort — over your head, over two chairs, whatever stands up. Yes, you'll feel ridiculous. The audio does not care.
- A parked car, engine off. Seats and headliner make it shockingly dead — just not in rain or on a busy street.
Avoid bathrooms and kitchens; tile is an echo machine. Then hunt the hum — fridge, fan, laptop, HVAC — switch off what you can and record at a quiet hour. A dead room plus a quiet background is 80% of "professional sounding," and it costs nothing.
How to read the script so it sounds said, not read
Your first take probably came out flat, careful, slightly scared — almost everyone's does. The fix isn't acting talent; it's changing who you're talking to. Talk to one person: before you hit record, picture one specific friend and explain the product to them.
In practice:
- Say the script out loud a few times before recording, and rewrite any line you keep stumbling on. If your mouth won't say it, it's written language, not spoken.
- Smile on the product line. You can genuinely hear a smile in a voice, and brands clock it even when they can't name it.
- Keep the small human stumbles. A tiny restart or an imperfect "so, yeah—" reads as real. Retake actual mistakes; keep the texture.
- Hold conversational pace. In 9:16, captions carry half the comprehension load, so you don't need slow announcer diction — talk at your normal speed and tighten gaps in the edit instead.
This is the same muscle as sounding natural on camera — minus the eye contact, which honestly makes it easier.
How to sync voiceover to b-roll so it feels edited, not pasted
Lay the voiceover down first and cut the footage to it — VO is the skeleton, b-roll is the skin. Pasting a voice track over an already-cut video is why so many VO versions feel off-beat.
The edit pass, in order:
- Drop your best full take on the timeline before any clips.
- Trim the dead air and the big gasping breaths between sentences. Leave the small breaths — a voice with zero breaths reads as robotic.
- Cut b-roll on sentence beats. New sentence or new clause, new shot. Cuts that land mid-word are what "pasted" sounds like.
- Land the product on its name. The moment the VO says the product name, the product is on screen — nudge clips until it's exact.
- Auto-caption, then fix the brand's spelling by hand.
- Duck any music well under the voice. If you notice the song during a sentence, it's too loud.
One hard rule: never pitch-shift or speed-change the voice to make it fit. A stretched voice reads as AI in about two seconds. If the VO runs long, tighten gaps or re-record the one line.
Should voiceover versions cost extra?
Yes — when they're an extra deliverable. A VO-over-b-roll cut is new recorded audio plus a new edit, not an export button, so it gets priced like work.
If the brief listed one talking-head video and the brand later asks for a voiceover version too, that's scope added after the price — quote it. A common approach is to price an additional VO cut of the same concept at roughly 25–50% of the base video rate, and a standalone voiceover video as its own full-rate deliverable.
Two quiet things to check before you quote:
- The deliverables line. If "1 talking-head + 1 VO version" was in the brief you accepted, it's already in the price you accepted. Honor it — and negotiate the next brief with that line in mind.
- Voice usage. Your voice is licensable the same way your face is. If a contract mentions cloning your voice or using it for AI-generated ads, that's a separate usage-rights conversation, not something that rides along free.
And if you'd rather be behind the mic than in front of the lens entirely, faceless UGC is a whole lane built on exactly this skill.
FAQ
How many words fit in a 30-second voiceover?
About 65–75 at conversational pace. Write 70, read it out loud with a timer, and cut the clause you trip on — there's always one.
Can I record straight into CapCut instead of Voice Memos?
You can — CapCut's voiceover tool records right on the timeline, which makes syncing easy. Voice Memos gives you a cleaner standalone file you can reuse when a brand wants three versions. A good combo: scratch take in CapCut, final in Voice Memos.
Can I just use an AI voice?
If the brand booked your voiceover, no — your actual human read is the product, and viewers are getting sharp at spotting synthetic voices. If a brand only wants an AI voice, they don't particularly need a creator — which tells you what that brief is really buying.
Why does my recorded voice sound so weird to me?
Bone conduction — you hear your own voice lower and warmer than everyone else does, so recordings sound thin to you — and normal to everyone else. Most people dislike their recorded voice at first, and it fades with practice. Ship it anyway.