Today I was once again the voice of a Norwegian oil worker. It is a role I seem to return to every so often for a repeat client, and it is always for an offshore safety video. But while I was recording the part today, I found myself thinking again about just how far audio tools have come. Because one of the most jaw-dropping things I have seen AI do is not write text, create images, or answer questions. It is what it can do to imperfect audio. To give myself a little authority on the subject, I have recorded more than 2,000 voice-led projects, with over 2,200 completed orders on Fiverr alone, most of them centred around voice over work, alongside a smaller number of video and website projects. So whilst AI has undoubtedly cost the voice over industry enormously, and damaged opportunities for many talented people who have worked hard to learn their craft, there is also one area where it genuinely helps those of us who are still lucky enough to be doing the work.

For years, getting good audio when you were away from a proper setup often meant improvising. I have recorded voice overs in all sorts of places and had to create a temporary recording space out of whatever was around me: a hotel room, a duvet, cushions, towels, anything that could help tame the room and take the hard edge off the sound. If the job needed done, you found a way to make the space work. That is why tools like Adobe Podcast’s audio cleaner are so impressive. What would once have been written off as awkward, boxy, echoey or just not quite good enough can now be improved dramatically in a matter of moments. Not always perfectly, and not without limits, but often to a level that is remarkably usable. By imperfect audio, I mean all the things that get in the way of a clean recording: a room with too much echo, a voice recorded too far from the microphone, a laptop mic, a phone recording, background noise, or a space that sounds hard, thin or hollow. Audio that is perfectly understandable, but just does not sound polished enough to hand over with confidence. A few years ago, a lot of that would have meant either re-recording it or spending far longer trying to rescue it. Now, with the right tool, you can take something that is clearly less than ideal and move it much closer to something usable, sometimes surprisingly so. That does not mean it replaces good technique, turns every rough recording into a perfect studio take, or removes the need for performance and delivery. But it does mean the gap between a difficult recording and a usable one is now much smaller than it used to be.

My own process with it is fairly simple. If I have a piece of audio that is less than ideal, I will often run it through Adobe Podcast first and hear what it can pull back. Sometimes the difference is subtle. Sometimes it is genuinely startling. From there, I am not just pressing export and calling it done. I still listen properly, tidy things up, and always take it into Pro Tools or Adobe Audition for further work afterwards. That might mean adjusting levels, taking care of breaths, smoothing edits, or just making sure the final result sits where I want it to sit. So this is not really about AI replacing the rest of the process. It is about AI giving you a much stronger starting point than you might have had before. And when you have spent enough time working with voice and audio, you realise just how valuable that can be. Of course, it is not magic. A trained ear can still hear the difference between audio captured in a properly treated space with a good microphone, and audio that has been rescued afterwards. But that is not really the point. The point is just how close these tools can now get imperfect audio to something polished, listenable and genuinely usable. You do not need to take Adobe’s word for it, and you do not need to read a list of features to understand the appeal. If you work with audio, even occasionally, the value of a tool like this becomes obvious very quickly. In fact, the change has been so significant that I eventually dismantled the recording booth setup I once relied on. That would have been hard to imagine a few years ago. Not because microphones, acoustic treatment and proper recording practice no longer matter, because they absolutely do, but because the technology around audio cleanup has moved on so dramatically that the old rules are no longer quite as fixed as they once were. That said, there is still a line. AI is clever, in some areas astonishingly so, but it still cannot truly replace an experienced voice over. Many people think it can. Many businesses think it can. But in my opinion, it still cannot convincingly replicate the judgement, timing, warmth, control and interpretation that come from someone who has spent decades behind a microphone. Cleaning up imperfect audio is one thing. Delivering a script properly is something else entirely. That is where experience still matters, and always will. For me, that is where the balance now sits. AI can help shape the sound, but it still cannot replace a real voice with years of experience behind the voice.

Find out what those dials can do to your audio!  https://podcast.adobe.com/en/enhance  when you forget to switch on the lapel mic and need to use the audio direct from the camera, this might just save you.  (so a friend told me – lol )