A video shot on an iPhone carries its own sound inside the file. When a fan, traffic or a fridge got into the take, the clip sounds wrong before anyone can say why, and no amount of colour correction fixes it.
One part of this is now fixable on the phone itself, and the rest still needs a computer. This guide covers both: what iOS does to a noisy clip on its own, the two conditions it needs, and the honest hand-off to a desktop app when your phone or your recording does not meet them.
What the iPhone can do on its own
The Photos app has an audio mixer for video, called Audio Mix. Open the clip, tap the edit controls, then tap Audio Mix, and you get four choices:
- Standard plays the audio exactly as you recorded it.
- In-Frame reduces sounds and voices from things that are not in the picture.
- Studio reduces background sound and reverb, so the voice sounds closer to a studio take.
- Cinematic puts the voices on a front-facing track and leaves the room sound around them.
Under those four choices there is a slider, and it raises or lowers the effect and the background noise. The slider is the part worth thinking about. Studio at a low setting often sounds better than Studio at the top, because the top takes the room out of the gaps between sentences too, and the voice is left sitting in silence.
The rest of the Photos tools change the picture or the timing. You can trim where a clip starts and stops, change the playback speed, and set which part of a slow motion clip plays slowly. None of those touch the noise. Audio Mix is the only one that does.
The two conditions it needs
Audio Mix is not on every video. Apple’s iPhone guide puts it plainly: it is there when you record video with Spatial Audio, on an iPhone 16 or later. On iPhone 18 Pro models it can also isolate music, separately from voices.
So both of these have to be true:
- The phone is an iPhone 16 or newer.
- The video was shot with Spatial Audio turned on.
That is set before you shoot, in Settings > Camera > Record Sound, where the choices are Spatial Audio, Stereo and Mono. If your clip was recorded in Stereo or Mono, or on an iPhone 15 or older, the mixer is not there, and nothing in the editor brings it back afterwards.
Where the phone stops
Audio Mix is a mixer, not a repair tool. Four things it does not do:
- It works on the mix, not on a noise source. A fridge close to the phone, or traffic on a busy road, still sits under the voice at the lowest setting. It is reduced, not removed.
- It cannot be applied after the fact. The recording settings are not retroactive. A clip from an iPhone 15 is a clip from an iPhone 15.
- It does not set a level. It changes how much room you hear, not how loud the file is for a podcast feed, YouTube or a client. That is a separate job, and the LUFS explainer covers the numbers.
- It cannot recover what was never recorded. A word buried under a gust of wind, or a take where the microphone overloaded, is gone.
Fix it before you record
Two minutes of setup is worth more than any amount of cleanup. On an iPhone 16 or later, one part is already automatic: when you record video with either Spatial Audio or Stereo, wind noise is reduced for you, and you can turn that off under Settings > Camera > Record Sound.
- Get close. Arm’s length from the speaker, not across the room. Phone microphones record everything in the room at roughly the same level, so distance is the only control you have over the balance.
- Do not cover the microphones. Cases, fingers and a hand wrapped around the bottom edge all block them. The microphones sit on the bottom edge and near the camera.
- Turn the room’s loudest thing off. A fan, an extractor hood, a projector. That single switch beats every setting.
- For a talk or a lecture, set the phone down and leave it. Moving the phone mid sentence adds handling noise. If that is your job, the lecture recording guide covers the room side of it.
Clean the audio on a computer
When the phone cannot do it, the fix is a desktop app, and it does not require exporting an audio track first. Since version 2.0, every tool in Gleam Audio takes video files for their sound: mp4, mov, mkv, webm and avi. The picture is never touched, and the result is audio only, saved as m4a, or in whatever format you pick in Convert.
- Get the file off the phone. AirDrop it to a Mac, import it over a cable, or put it in iCloud Drive or Files and download it. A long clip is a large file, so give it time.
- Drop the .MOV on the Gleam Audio window, or click Open Files…. No extraction step, no converter in between.
- Pick Remove noise and set Strength. Start low and raise it only until the gaps between sentences go quiet. Too much strength is what makes a cleaned voice sound robotic.
- Use the Before and After switch to hear the original and the result from the same moment. Since version 2.1 this works on finished files, so you can compare two Strength settings without leaving the window.
Gleam Audio runs on macOS, Windows and Linux, does the work on your own computer, and is free for personal use up to 20 files a week with every tool. A license covers work, from $10. Nothing is uploaded, which matters for a client’s footage.
Put the sound back on the picture
The result of the cleanup is audio, not video, so the last step is putting it back. Two ways, and both keep the picture untouched.
In a video editor. Mute or delete the clip’s original audio, drop the cleaned file on a track below it, and line the two up. The cleaned file keeps the original’s length, so it sits where the old sound sat.
From a terminal, without re-encoding the picture:
ffmpeg -i IMG_0001.MOV -i IMG_0001-clean.m4a -map 0:v:0 -map 1:a:0 \ -c:v copy -c:a copy -shortest IMG_0001-final.mp4-map 0:v:0 takes the video stream from the original and -map 1:a:0 takes the cleaned audio. -c:v copy copies the picture without touching it, so nothing is recompressed. -c:a copy keeps the m4a audio as it is, which avoids a second lossy pass. -shortest ends the file where the shorter stream ends, so you do not get a tail of silence.
Then check the join. Play a second near the start with a lip flap or a hand clap, and one near the end. A cleaned file that was trimmed anywhere will drift.
If you want the opposite, a video with no audio at all, the remove audio from a video guide is the one to read.
When it won’t help
Some recordings cannot be saved, and it is cheaper to know before you spend an hour on them.
- Noise as loud as the voice. If the fan is at the same level as the speaker, removing it removes the voice with it.
- A gust, a bang or a chair scrape over a word. Noise that lasts a fraction of a second and lands on a syllable has nothing left to reconstruct.
- A clipped take. When the input overloaded, the waveform is flat at the top and the detail is gone. No tool invents it.
- Music that has to stay. The noise remover is tuned for speech, and on music it takes out more than you want. Gleam’s own docs note this under known limitations.
If your take is one of those, record it again. That is the honest answer, and it is usually faster than the cleanup.
The short version
On an iPhone 16 or later, shot with Spatial Audio, try Audio Mix in Photos and set Studio with the slider low. On anything else, move the clip to a computer, drop it straight into Gleam Audio with no extraction step, run Remove noise at a low Strength, and put the cleaned audio back on the picture.