Which tool to choose
To clean up spoken voice, the go-to tool is Adobe Podcast Enhance: free, no Adobe subscription, designed precisely to turn a home recording into clean audio. It doesn't require an account for basic use and accepts files up to about half an hour.
If you need to process very long files, in bulk, or to work on the audio of a video, that's where the paid plan or a more complete audio editing tool comes in. For the common case (a voice note, an interview, a short episode) the free version is enough.
How to do it
From browser or app, the path doesn't change.
- Start from the least compromised file you have. If you recorded in several ways, choose the one with the least noise: the AI improves, it doesn't recreate missing audio.
- Upload to the tool. Open Adobe Podcast Enhance, upload the file (respect the free plan's duration and size limit), wait for the processing.
- Listen to the before and after. The tool shows the comparison. Check that the voice is clean but still natural, not metallic.
- Adjust the intensity if you can. When an enhancement intensity slider is available, don't push it to the maximum: past a certain threshold the voice becomes robotic. Look for the point where the noise disappears but the voice stays human.
- Download the clean file. Export it and listen to it again on the device where it will be used (headphones, phone speakers): audio that sounds fine on headphones can sound different elsewhere.
There's no prompt to copy here: these tools work on the file, not on a text instruction. The control you have is the intensity slider and the choice of starting file.
A concrete example
Sara interviewed a craftswoman in her workshop with her phone. The recording has the hum of a fan and the echo of the empty room. She uploads the twenty minutes to Adobe Podcast Enhance. The first result is very clean but the craftswoman's voice sounds artificial. Sara redoes the processing, lowering the intensity: the hum is still nearly gone, the echo greatly reduced, and the voice goes back to being warm. In ten minutes the interview is publishable, without having recorded again.
When it does NOT work (and how to fix it)
If the voice comes out metallic or robotic
It's the most common side effect of aggressive enhancement. Fix: lower the filter intensity and reprocess. If the tool doesn't have the slider, try a second, less aggressive tool: sometimes a lighter improvement sounds better than a perfect but fake one.
If there are two overlapping voices on the same track
These tools struggle to separate voices that overlap on the same microphone. Fix: when recording, give each person a separate microphone or device. If it's already done, the only real solution is to manually cut the overlap points: the AI doesn't untangle them reliably.
If the noise is louder than the voice
When the disturbance covers the voice, there's no information to recover. Fix: no tool invents words lost in the noise. If the voice is still intelligible under the noise, the AI pulls it out; if it's buried, it has to be recorded again. Better to prevent: bring the microphone close to the mouth, turn off fans and close windows.
If the file is too long for the free plan
There's a duration and size cap. Fix: split the recording into several files under the limit, process them one by one, and recombine them in a free audio editor. Or consider the paid plan if it happens to you often.
A tip from someone who actually uses it
Clean audio matters more than a perfect image. Someone watching a video forgives a mediocre shot, but closes everything if the voice is noisy or reverberant. Before tending to anything else, fix the sound. And remember that five minutes of care during recording (microphone close, a room with curtains and rugs that dampen the echo) are worth more than any cleanup afterward.
Frequently asked questions
Can I remove the background music and keep only the voice?
In part. Some tools separate voice and music, but if the music is loud and continuous the result is imperfect. For clean control, record the voice without music and add it afterward in an editor.
Does it work on a video's audio too?
The basic free tool works on audio files; for videos the support is often in the paid plan. Alternatively, extract the audio track from the video with a free editor, clean it up, and put it back in the video.
Does it also improve an old or low-quality recording?
It can clean up noise and reverb, but it doesn't increase the fidelity of a recording that was poor to begin with. It removes the disturbance; it doesn't add detail that was never captured.
Can the AI make any noise disappear completely?
No, and it's better not to expect it. Pushing the filter to "zero noise" sacrifices the naturalness of the voice: the result is clean but false. A bit of residual ambience makes the audio believable. The goal isn't absolute silence, it's a clear voice that sounds like a real person in a real room.