Karaoke - Deep Dive AI Podcast¶
This podcast was created using Google's Notebook LM
I feed in my Karaoke Article and this was the results.
Feedback¶
Pros¶
- The speakers are lively and engaging.
- Interesting conclusions
- Good story telling
- Unique perspective on the original article, which provides some feedback to me, the author, for how my readers could interpret the post.
Cons¶
- Misinterpreted the complications I had with the "middle" channel
- Wrong conclusion about the contest. I did share he got a silver medal.
Text Transcription (provided by AssemblyAI)¶
Alex: Ever been listening to music and think, man, I wish I could take the lead singer out of this so I could sing it at karaoke night.
Pat: Happens all the time.
Alex: Well, that's what we're diving into today. We found this blog post by a self proclaimed, get this, recovering ten x developer, okay. Who used his coding powers to, well, take karaoke night to a whole new level.
Pat: Wow.
Alex: Helping his son prep for a singing competition using AI. The blog post is called Karaoke y. Scott. W h y. Catchy.
Pat: I like it.
Alex: So Scott son has this competition. They need to isolate vocals from these pre recorded tracks.
Pat: Okay.
Alex: And, I mean, I guess most people would just try to find a karaoke version, right?
Pat: Or use some audio software, you know, try to minimize the vocals.
Alex: Yeah, just like reduce the vocal track or something.
Pat: Exactly.
Alex: But not Scott. This guy goes all out.
Pat: Oh, I see this coming. I can tell.
Alex: He's like, if it's worth doing, it's worth overdoing.
Pat: Love it.
Alex: Which I thought was hilarious. So he decides to, you know what? I am going to build a whole karaoke system from scratch using AI, of course.
Pat: Of course. Why not? Why not?
Alex: And it's interesting because it is a good example of why those simpler solutions, they don't always work, especially for something. If you want to be serious about it, if you have a vocalist who actually wants to use this to train.
Pat: Exactly. And those traditional methods, they kind of rely on just removing certain frequencies, which for casual singing along, it's fine. But if you really want to hear yourself, it just makes everything muddy. It's not good.
Alex: So Scott jumps into this whole world of AI music source separation with this tool called splitter. Have you heard of this?
Pat: Yeah. So spleeter uses a neural network, basically, it's trained on tons and tons of songs to learn to identify and separate all the different parts of the music.
Alex: Okay.
Pat: To vocals, instruments, even drums.
Alex: Wow. So we're talking about Scott, who's a coder, not like an audio engineer or anything, right? Diving into collab, notebooks, APIs, static site building, all to just, you know, wow, give his son a shot.
Pat: That's the part I think is so fascinating. Right?
Alex: Yeah.
Pat: Not even just the tech, but that he was willing to just go for it. Like it's such a specific and kind of personal goal.
Alex: Totally.
Pat: I don't know, it just shows you how accessible these tools are becoming. That anyone can use them.
Alex: Yeah. And that you can use them for such, you know, exactly. Weird, specific things.
Pat: Totally.
Alex: Okay. But like, anything involving AI, it's not.
Pat: Always, you know, it's not gonna be easy.
Alex: Smooth sailing. And Scott's actually really honest about, like, the challenges he ran into.
Pat: Of course. Yeah. I mean, it can be amazing, this AI stuff, but it still needs that human touch.
Alex: Yeah.
Pat: You know, especially if you want something that sounds really polished.
Alex: Right. It can't do everything for you.
Pat: Not yet, at least.
Alex: Yeah. One of the things that tripped him up was this whole thing, this middle channel.
Pat: Oh, yeah.
Alex: What is that? And why did it, like, totally derail his whole AI karaoke machine?
Pat: Okay, so the middle channel. Imagine listening to a song, right? Like, all the instruments and the vocals, they're kind of laid out in front of you, almost like a 3d map of the sound.
Alex: Okay.
Pat: And the middle channel is that center part. It's where you'll find a lot of the stuff that blends everything together.
Alex: Okay.
Pat: It's not just about, like, panning things to the center. It's reverb vocal harmonies, instruments that are kind of meant to glue the whole mix together.
Alex: I'm picturing it.
Pat: You might not really notice them on their own, but they really add to the overall feel. And that's where these AI tools, like splitters, that's where they can sometimes struggle. Cause it all gets a little intertwined with the main vocals in that middle channel.
Alex: So it's like you baked a cake, and now you're trying to take out the sugar.
Pat: Yeah, exactly. Good luck with that.
Alex: Yep.
Pat: So that's what happened with Scott. The AI was doing its thing, doing great, but it really needed a human to come in and fine tune to make sure those vocals were, you know, crystal clear. Especially for a competition.
Alex: Yeah, for a competition, you need it to be perfect.
Pat: Right.
Alex: And he actually says in the blog that the final product was, like, 80% AI generated. 20% human stitched.
Pat: Yeah.
Alex: Which, I don't know, that just seems like AI these days, you know?
Pat: Totally.
Alex: It can do a lot, but it still needs us.
Pat: Yeah. It's like it can do the heavy lifting, but we still need to, you know, kind of guide it.
Alex: It's like when I'm prepping a presentation.
Pat: Okay.
Alex: Like, I might use AI to help me get my thoughts down.
Pat: Yeah.
Alex: Maybe even design some slides or whatever. But it's still me, you know? It's my voice.
Pat: Exactly.
Alex: My perspective that makes it all work together.
Pat: It's that human element, not just in the final product, but in the whole process.
Alex: Yeah.
Pat: Think about Scott. I mean, diving into all that new tech. He wouldn't have learned all that if he just downloaded a karaoke track.
Alex: That's a really good point. So, okay, all this tech talk, did it actually work? Did his son nail the audition? Was it a success?
Pat: Well, that's the thing. He doesn't actually say if he won or not.
Alex: Oh, really?
Pat: He kind of leaves us hanging, so.
Alex: We don't know if it was a standing. After all that work, he really focuses.
Pat: More on the journey, what he learned along the way.
Alex: Interesting.
Pat: I think it says a lot about how the project was about more than just winning. It was about supporting his kid, doing something cool with technology, maybe learning something new himself.
Alex: Yeah, yeah, for sure.
Pat: And speaking of which, his big takeaway from all this is 99% practice, 1% technology, which I love, because it's so easy to get caught up in the fancy new thing.
Alex: Totally.
Pat: But at the end of the day, whether you're singing or coding or, you know, podcasting, it's all about putting in the time, the work.
Alex: He even jokes that next time, instead of building a whole karaoke system, he'll just sneak into his son's teacher's suitcase and record practice directly. Well, there you go, hiding in a suitcase. I mean, talk about extreme lengths for your kid's karaoke career.
Pat: That's dedication for you. Maybe a little over the top.
Alex: He's got a good sense of humor about it, though.
Pat: Definitely.
Alex: Yeah.
Pat: But it's a cool story, right?
Alex: Yeah, it really shows how technology can bring families together. You know, like how many dads would build a whole AI karaoke system just for their kid? It's pretty sweet.
Pat: Totally. And it's easy to focus on the big, impressive AI stuff, but I think it's these smaller stories that are more interesting.
Alex: Yeah.
Pat: More personal.
Alex: It's not even just about the end result, you know?
Pat: No, exactly.
Alex: Like all that learning and problem solving, the whole process can be just as important.
Pat: I think it goes back to what we were saying about AI and human input, that 80 20 split. AI isn't going to replace us. It's more like it helps us do things we couldn't do before.
Alex: And it's like Scott said, 99% practice, 1% technology. You still got to put in a work.
Pat: It's a good reminder for all of us.
Alex: It is. So I think this deep dive taught us a lot more than how to make a karaoke track. It's about using technology in creative ways.
Pat: Definitely, and not being afraid to experiment, even if it seems impossible.
Alex: And who knows? Maybe it even inspired some people listening to, I don't know, dust off their old karaoke machines or maybe build their own or build their own. There you go.
Pat: That's a wrap.