A few weeks ago, I shared a video about my little AI adventure with Google NotebookLM. I had it create a sweet little story about rabbits and good luck. And honestly? I was impressed.
It put together a cute video in just a few minutes. But there was one problem. The voice.
It wasn’t bad. It was just… a bit much.
I’m sure you’ve heard the voice before. It’s the high-pitched, fast-talking, overly excited voice. At my age, my ears are not especially fond of that style. I prefer something a little calmer. Slower. Less caffeinated.
At the time, I let it slide. I was new to all of this and figured, “Well, that’s just how AI voices sound.” But it got me wondering. Is there a way to change the voice on the NotebookLM video?
Turning It Into a Personal Challenge
I could have tried prompting NotebookLM to adjust the voice. But instead of going down that rabbit hole (pun intended), I decided to try something different. Instead, I wanted to see if I could replace the AI voice with my own voice.
No, I didn’t want to hear myself talk. I just wanted to do it to see if it could be done. So I challenged myself, and I accepted my challenge! So here’s what I did.
Step 1: Extract the Audio
To do this, I used the LumaFusion video editor. Any video editor would do, but since I’m most familiar with LumaFusion, I chose it as my audio extraction tool. This is how I did it:
- I added the original rabbit video to the video timeline
- I extracted the audio
- I exported the audio as an .mp3 file. That file became the “script” I would upload into the voice cloning tools.
Step 2: Cloning My Voice with ElevenLabs
At first, I used ElevenLabs. If you’re not familiar with it, ElevenLabs is known for its AI-generated voices and text-to-speech capabilities. I had already used it for a talking-head video, so I was familiar with it. I wasn’t too crazy about the sound of the cloned voice, but since it was created from just a 10-second voice sample, I couldn’t complain too much. The voice sort of sounded like me.
Next, I uploaded the .mp3 script, selected my cloned voice, and let it process the audio. When it was done, I downloaded the file, opened LumaFusion, and replaced the original high-pitched voice with my own. That’s all there was to it, or so I thought!
Step 3: Discovering Fish Audio
Opened on my desktop was an Apple Notes document. In that document, I had written a note to myself that said, “Investigate Fish Audio.” I don’t remember when I wrote that note, but it was staring me in the face so, I figured now is as good a time as any to check it out. By the way, Fish Audio is another voice cloning platform.
I went over to Fish Audio and gave it a 10-second sample of my voice. The difference this time was that the voice quality was much better. The likeness was noticeably better. It felt more like me. The pacing, the tone, and the overall sound were impressive. So naturally, I had to upload my .mp3 script to Fish Audio to see what it could do.
Fish Audio processed the file, I downloaded it, and swapped it into the video. The Fish Audio file became the final version.
Cloned Voice: Was it Worth It?
Yes, it was worth it. Not because the world needed to hear a rabbit story narrated in my voice. But because I proved to myself that I could do it! And that’s what this Low-Tech Grandma journey is about. Trying things, testing tools, and not assuming something is “too technical” before giving it a try.
The cloning process itself wasn’t complicated. I’m just happy that I decided to give Fish Audio a try. But don’t take my word for it. Try it yourself. Actually, watch it yourself. I created a video of my experience.
I invite you to watch the full video here: