Clone a voice? Simples!
A video editor friend of mine related a problem he had recently where the person being interviewed for a video insert said 2002 instead of 2020, so he found an online site where you could uploaded a sample of the voice, type in what you actually wanted it to say and it generated the correct phrase in the correct voice for you using AI.
It was a snippet of information I logged away and I bookmarked the site he mentioned. Come forward just a few days and I’d edited the video from 2 GoPro cameras for a good friend of mine for his YouTube Channel about extreme narrowboating (www.wyeinvader.uk). I asked him to come round, watch the video and record a voiceover that I could drop in, it all went well, I edited the V/O in and uploaded the draft video for him to approve.
He came back to me after watching the video and approved the video apart from a tiny one word error - he’d said Symonds Yat but meant to say Sharpness, as there were technical and safety details about tides in the video we needed to get it correct. My immediate thought was to find the word somewhere else in the video and copy/paste it in, I then remembered the conversation with my editor friend about using AI......
I opened the site (play.ht/), it asked for a sample of the voice I wanted to use and I uploaded 30 seconds from a section of the video, I then typed in the correct phrase and within 60 seconds I had an audio file that sounded like Frank saying ‘the tide is still running at Sharpness’.
As the actual word could have two meanings (a place or a description) the inflection was just slightly off but good enough for the video. You can hear the 2 clips below.
I can fully understand why some people are now so opposed to AI, yes it can have huge benefits that was previously only possible by science fiction writers but it could also be used for nefarious purposes - who, if anyone makes the decision on control?
Original file
AI generated file