Are voiceless podcasts the way of the future?

 
 
 
 
 

If you’ve ever spent time on YouTube, you’ve probably come across “faceless” channels. 

Some even employ AI voices, allowing them to focus purely on content creation without involving their personal identities. This style of creation is becoming increasingly popular for video because it involves far less work. You don’t need a video camera, lighting or recording space. I can see the appeal to some creators.

It made me wonder: could podcasts follow a similar trend and become entirely voiceless? Let’s be clear here as to what “voiceless” means in this context. There is still technically a voice, but it’s not your human voice. 

With the increasing popularity of cloning voices using AI, are we on the precipice of stepping into a world where synthetic voices are the norm?

Imagine using AI to replace your voice altogether. No more recording sessions, just write a script, or you could go even further and have AI write your script for you. Then let an AI-generated voice say it all and bam, you’ve got yourself a “podcast”. 

To some, this might seem like a utopian shortcut—minimal effort for maximum output. To me, it initially felt very wrong and entirely impersonal. But, in the spirit of experimentation, I decided to give it a go and share my findings with you.

Steps in the Experiment

I used an AI voice cloning tool that is available inside of Descript. The process was fascinatingly swift and a bit eerie. Training the AI on my voice took about 20 seconds; then I challenged it further by having it generate a script on podcast launching. Let’s be clear: I would never typically outsource my scriptwriting to an AI. This was purely an experiment. 

The AI churned out a passable, albeit lifeless, script in about 20 seconds, and then I had it read the script in my AI-trained voice.

First Impressions

Listening to the AI version of "me" was surreal. The voice carried fragments of my character but was undoubtedly robotic. It lacked the emotion, cadence, and personal connection integral to podcasting. Spoiler alert: AI Brianna was boring. It was a stark reminder that the human element in podcasting is irreplaceable—for now, at least.

Potential Use Cases for AI in Podcasts

Despite the shortcomings, there are potential uses for AI voices. For instance, minor corrections—like fixing mispronounced names—could be efficiently addressed without setting up all the equipment again. While entire segments might still sound off, minimal edits might go unnoticed, saving time and effort. Yet, this is only feasible if creators have their AI voices pre-trained, which raises ethical and legal questions that I’d be interested in diving into further at some stage.

The Real Value in Human Connection

The crux of podcasting is creating a personal connection with listeners. As I recorded my thoughts, reflecting on the AI tool's output, it was evident that a monotonous, emotionless voice couldn’t forge the same bond. Most listeners seek authentic voices that convey real emotions, thoughts, and nuanced insights. AI, however advanced, presently fails to encapsulate that essence. The jokes don’t land, pauses are non-existent and I found myself unable to even digest the information that the AI voice was providing.

My little experiment with AI voices was illuminating, if not slightly unnerving. Although AI might handle statistical or factual content delivery, the soul of a podcast lies in its host's unique voice. This experience made me appreciate even more the art of podcasting and the irreplaceable value of human connection in our auditory world.

So, what do you think? Would you listen to an AI-generated podcast, or do you crave the genuine connection brought by a real voice? 

 

Transcript:

  • [00:00:00] Brianna: You know how everyone's talking about AI these days, and everyone's scared of it, or they love it. Everyone's got an opinion. Well, I want to have an opinion too, but specifically for podcasters. Today, I'm going to have a little look through the AI feature within Descript, which claims to be able to clone your voice, meaning that you could say actually nothing.

    [00:00:23] You could get Descript to write your script, then have your AI voice. Say the script and then publish it without you actually doing any speaking at all. I'm very dubious on this as to whether it will do as intended and whether you'll be able to tell the difference as to whether it's the AI version of me or the real version of me, so let's put it to the test.

    [00:00:51] The first thing we're going to do is we're going to go up to the AI speakers section here of Descript. And I've got no speakers in there. I've got nothing going on. We're going to go new speaker and we're going to add myself. There I am already. Confirm. Now we need to record a statement to train the speaker.

    [00:01:09] Okay, here we go. I want Descript to create an artificial.

    [00:01:16] Okay. So I've recorded my thing. thing that it told me to record. And now I am going to submit. I have to agree to the fact that yes, I'm authorizing the script to use it. So we submit, and now I'm going to set a timer to see how long it takes. Oh, it didn't, it took, that took about 30 seconds for it to authorize, and it says now you can use Brianna for speech generation.

    [00:01:42] Okay, I'm actually a little bit weirded out by that because that's my flippin voice. Now I'm going to go to a completely new session, so bear with me.

    [00:01:55] And we're going to go in and we're going to start a new audio project. So that [00:02:00] is going to be no video attached. At this point, what I'm going to do is I'm going to start writing. So I'm going to go here and then I'm going to actually go up over to the right here and get the underlord to help me write some scripting.

    [00:02:15] So if we go over here to write a script, okay, and then Give Underlord a topic and a style and get a first draft. So we want starting point to record a podcast episode about how to start a podcast. Tone should be conversational, humorous,

    [00:02:46] but also professional. Audience should be first time podcasters. Demographic between, let's say 20, 20 and 50.

    [00:03:03] Length should be, let's say. 800. I don't actually know how long is too long. Let's say 800 words. Okay, so based on that, I'm just now going to hit submit. I've given it some idea. And now, uh, we're going to get the overlord to give us a bit of a script. We're going to wait and see what it comes in with. Okay, and then I'm going to go add.

    [00:03:30] to script Okay, and then now that it's in here. I'm going to go add speaker and I'm going to use my AI voice Oh, what and now I'm going to wait and let's see how long it takes for this all to become Actual AI voice Brianna so far. I have done nothing other than Decide to tell Descript what to write [00:04:00] a podcast episode about.

    [00:04:01] This is a podcast episode that I have done before. In fact, I had a whole course that was called let's get launchy. So it was about how to start a podcast. In fact, I do consults where I help people through the whole process of starting a podcast, what they're going to need to know, the kind of gear recommendations, all of that stuff.

    [00:04:17] So it feels like a good one to do a podcast episode on that, uh, feels pretty general. I could definitely have given it more. tips, more things to talk about. And knowing now what I know about actual length, 800 words is nowhere near enough. I'd like to see it be expanded. And maybe I could even do that here.

    [00:04:35] One of the real tests as well would be if I did put this on the pump up your pod podcast and not tell anyone. that it was an AI version of me and see if anyone said anything. I'd feel kind of weird about that. All right, so it now looks like there is everything there. The thing I can see straight out before I even listen to it is that this waveform is way too consistent.

    [00:05:00] In fact, it looks almost the same. In a voice that isn't AI generated, uh, it normally has a lot more kind of undulation in the waveform. So it'll have things that are much louder, much bigger, like waveform looks a lot bigger and then smaller, and then there's going to be filler words and things in normal text.

    [00:05:17] Whereas this, a normal speech, whereas this is, is not that at all. Now, what I'm going to do is I'm just going to close this down and I'm going to get rid of the underlord and I'm going to put my headphones on so that I can hear what is actually going on. And that scares me. You're getting my actual reaction to this.

    [00:05:33] Okay. I've never heard any of this before, so let's give it a listen. Hey there, podcast enthusiasts. Welcome to our very first episode where we'll dive into the wonderful world of podcasting. If you're here, that means you're eager to start your very own podcast. And we couldn't be more excited to help you on this journey.

    [00:05:58] I love that it's the [00:06:00] most monotone, couldn't be excited ever. And we couldn't be more excited. You clearly could be a lot more excited. AI Brianna? First things first, why should you start a podcast? Well, if you're anything like me, you probably love talking, sharing stories, or maybe you have some expertise that the world just needs to hear.

    [00:06:21] Plus, it's a fantastic way to meet interesting people and build a community. So far, this is actually pretty good. I was expecting this to be a lot more American. There is actually inflection here in my voice that definitely sounds like me, which is a little bit weird, but I can also very much hear that there's no enthusiasm.

    [00:06:42] Okay, I'm a very enthusiastic speaker, you may have noticed that already, but there's nothing there. It's very much a monotone kind of situation and that's definitely what I've heard from other AI, certainly with podcasts that have just been, you know, AI voices. There's nothing there that shows me that they have actual life.

    [00:07:01] Okay, let's keep going. All right, so you've decided to start a podcast. Fantastic. Now comes the fun part picking your podcast theme. This part is crucial because you want to talk about something you're passionate about. Remember, if you're bored, your audience will be too. I'm bored. Now onto the gear. Don't worry, you don't need to break the bank to get started.

    [00:07:24] At the very least, you'll need a decent microphone, headphones, and a computer. There are plenty of budget friendly options out there. Just Google best budget podcasting gear, and you'll find loads of recommendations. Once you've got your theme and your gear. At this time, you can't give it the sparkle that is a personality that is a person.

    [00:07:46] Sounds like Daria. I don't know if you're of the generation where Daria existed, but there was this awesome show called Daria and her voice and the way she spoke was just unenthusiastic, she wasn't interested. She had this vibe about her [00:08:00] and that's what my takeaway is for this kind of thing, for AI.

    [00:08:03] voices is that, yeah, technically you can do it. Technically you can train it to sound like you, but there's a degree of personality that it definitely can't do right now. So let's keep going. Start conversations and most importantly, have fun with it. Remember, the most important thing is to enjoy the process.

    [00:08:27] Thanks for tuning in and happy podcasting. The way it says podcasting, it's kind of got a little bit of an Irish vibe to it in the actual accent there. I don't know if you can hear that, but podcast casting, it sounds a bit strange. So there's a few little like accent things. Sometimes. I'm a little bit more American than I am.

    [00:08:47] I mean, I'm not American at all. It did actually do a better job than I thought it would with the actual Australian accent, because I know that one can be a tricky one. It definitely got the memo on kind of the way I speak. There was more of those sparkly higher tones, which is something that my voice is known for.

    [00:09:06] There was absolutely no personality in it at all. It was boring to listen to. I actually find myself not being able to really listen to what the actual information is because there is no, uh, expression. I'm not getting the actual info out. I'm not learning how to start a podcast properly because I'm, I'm getting lost in just the, the minutiae of this.

    [00:09:31] It's a drone. It doesn't have a. Beginning, middle, and end. It doesn't have phrasing. It doesn't have pausing. You know, all that sort of stuff is, is very important. What you could do, I guess, if you were really wanting to have it be like a faceless, you know how they have faceless YouTube channels, uh, you could do like a faceless podcast or a voiceless, voiceless podcast.

    [00:09:51] You could do that. Do something like that and spend a little bit of time in the actual editing phase of something like this and creating more like drama, [00:10:00] drama, uh, so you could do things like creating extra pauses. Welcome to our very first episode where we'll dive into the wonderful world of podcasting.

    [00:10:10] If you're here. Oh, it gave me a breath. That is something I didn't hear the first time around. It actually put in a breath. Look, there it is. A little mini breath there. Podcasting. If you're here, Wow. That is interesting. Because, I mean, that's part of the thing, right? Is that if you, in a normal speech, you can hear someone breathing and if you can't hear them breathing, then that's very off putting and certainly in earlier versions of AI, I have not been able to hear breathing and that just puts you off immediately.

    [00:10:41] It just doesn't sound right. So the fact that they're putting artificial breaths in, It is quite interesting. So what I'm going to do while I'm over here in Descript as well, I'm going to go over to properties and I'm going to just put a few little effects. I'm actually going to try and make it sound a little bit better by changing up the EQ a little bit.

    [00:10:59] All right. And let's just have a little bit of a place where I'll listen along as I'm doing it. And we couldn't be more excited to help you on this journey. First things couldn't be more excited to help you on this journey. First things first, why should you start a podcast? Well, if you're anything like me, you probably love talking, sharing stories, or maybe you have some expertise that the world just needs to hear.

    [00:11:22] Plus, it's a fantastic way to meet interesting people and build a community. Alright, so you've decided to start a podcast. Alright! Now comes the fun part, picking your podcast theme. This part is crucial because you want to talk about something you're passionate about. Okay, so I gave it a bit, if you look at that EQ, I needed to give it a little bit more depth, so bottom end.

    [00:11:47] I've put, given myself a little bit of an extra boost here because I feel like I needed around the 200 hertz in this version of myself. And then I'm reducing this area, probably around the 500. [00:12:00] And then as we move up, it just sounds a little bit too sibilant. There's too much of that high end kind of area in this particular recording.

    [00:12:08] So to make it sound a little bit better, I'm just taking a little bit of that out. I'm going to listen to a little bit more now and maybe make a few more adjustments. Why should you start a podcast? Well, if you're anything like me, you probably love talking, sharing stories, or maybe you have some expertise that the world just needs to hear.

    [00:12:27] Plus, it's a fantastic way to meet interesting people and build a community. Fair. Right? So you've decided to start a podcast. Fantastic. That's probably as good as I'm going to get that as far as the EQ goes. Then I can now just give it a little bit of, it shouldn't need much, but a little bit of compression.

    [00:12:44] I'm going to bring it to a minus 13 at a split of ratio 3 to 1. Attack is fine, release, all that's all good. Okay, so we'll do that and then we'll hear the difference. Like me, you probably love talking, sharing stories, or maybe you have some expertise that the world just Bit more up here. It's a fantastic way to meet interesting people and build a community.

    [00:13:07] All right, so you've decided to start a podcast. Fantastic. Now comes the fun part, picking your podcast theme. Now comes the fun part, picking your podcast theme. So I can't give it any extra pizazz. There's no like click for extra excitement pizazz button here, but I can give it a little bit of that drama as I was talking about with giving it some extra space.

    [00:13:31] So I'm just going to split the track here. Why can't I split the track? Oh, I have to convert it to audio. Okay. That's weird. Okay, there you go. So I couldn't do anything with it because it wasn't technically audio. It's so weird. Okay, right click, split clip, or just press S. Uh, and then I'm just going to add some gap, but not too much gap.

    [00:13:57] And I'm going to try and get it to sound a little bit more of the same, [00:14:00] like of the pace you would expect. Fantastic. Now comes the fun part, picking your podcast theme. Yeah, so you see, but there's no hard T. Fun part, picking your podcast theme. Okay. Unpart picking your podcast theme. The other thing you could do is, if you were really going to town on this, is you could make, like, create a little bit of room tone.

    [00:14:21] So you can add some room tone here into script, which normally isn't great. But anyway, let's have a look. Unpart picking your podcast theme. This part is crucial because you want to talk about something you're passionate about. Remember, if you're bored, your audience will be too. Trust me. The breaths thing is weirding me out.

    [00:14:41] It's true. You're looking to be able to hear the breath. And so the fact that that's in there, it does sound more natural than I would expect it. It's kind of strange. Isn't it? What do you think? I would love your feedback. This is just a very quick initial little look at how you could even manipulate it a bit, create a little bit of extra drama, put a few more gaps in.

    [00:15:03] Uh, you would, you could mix some stuff with your EQ and compression so that it sounds a little bit more present, sounds a bit more natural. Well, not really natural. You can't really make it sound natural. There's nothing really else that I could suggest from an EQ or like in the actual area here. And if you get creative with anything, then that's just giving it sounds that don't make sense.

    [00:15:22] Maybe, I mean you could try giving it a little bit of a Reverb but that just means that the room tone like it'll make it sound like there's more going on in the room This part is crucial So if you want it to sound worse you can definitely add in That but you would never put them you would never put it in a mix of 80.

    [00:15:42] You'd put more like 10, 15, maybe, if you were wanting to put a reverb on it. This part is crucial because you want to talk about something. Yeah, I mean, I wouldn't use that. I'd use it, obviously, I use it in vocals, in singing, you put a reverb on, but for a voice, I mean, [00:16:00] two. This part is crucial because you want to talk about something.

    [00:16:04] Yeah, you could do that. It gives it that little bit more presence. So that's what I think. All right, so let's talk about it. This is interesting. This is a weird world that we're living in right now, where I can record a short script and then AI goes to town on making it sound like me. I don't think it did a great job.

    [00:16:25] Like, I don't know whether I'd be getting away with that, I guess, essentially, is what I'm saying. So after now doing that, I'm, I have to say, I'm pretty impressed with it overall, how it actually did. It doesn't make me scared, it makes me kind of, uh, Enthusiastic about the future about what potentially the use cases could be for I know that there's deep fakes And there's all sorts of things that are happening that are wrong that are bad that are definitely not what we want But as a fun little experiment for me and to see how it actually went Being me, I think that that was pretty cool.

    [00:16:59] I definitely don't think I'm going to be duping anyone by having a podcast episode on Pump Up Your Pod that is completely AI. They're used to the way I actually sound. I have more inflection, I have more dynamic range, I have more intrigue, I have more personality, I have more presence. The AI version of me does not have any of those things.

    [00:17:19] I guess the use case for it right now, especially if you used a script, is if you stuff up a word, you've recorded a section, but then you put your mic away and you don't want to bring it back out and you want to just re record or you want to just have like a few words replaced, you can do that. And if it's got your AI voice trained, then it's easy to sort of drop that in and it will do its best.

    [00:17:41] There's use case there for that sort of thing. Find a replace, make it sound better if you've stuffed up something. Okay. I think that as far as replacing yourself, making it like a, a voiceless channel, like they have faceless channels on YouTube, voiceless channel in, uh, audio. [00:18:00] I don't think we're there yet in any real way.

    [00:18:02] I was listening to a podcast. That I knew was an AI podcast a couple of weeks back, and I was so bored, so bored, I couldn't get to the end. Yeah, I clicked on it because the title was good, it was catchy enough, it got my attention, but it Definitely didn't hold my attention and it's not the kind of podcast I would then go back to because it gave me nothing.

    [00:18:25] The whole point of the audio experience is to have an experience, is to have something in your ears that you feel connected to, that you want to listen to. And if the AI is like this, I don't want to listen to that. That is boring. AF. At this point right now, I think it's fun. I think it's silly. I think there's a use case for it in a very small sort of section of the market, but to have a true experience is at this point in time, an actual human voice speaking with their actual dynamics and their true representation.

    [00:18:59] I hope you liked this. What do you think? Did I dupe you? I mean, I kind of duped you because you watched me doing it, but if you go and listen back to a little bit of that, how do you think it went? playing with my voice? Did it sound good enough? Do you think that if it was, you didn't have any context and you were just listening to it, do you think that you could just, you know, be okay with it?

    [00:19:20] That it would sound good enough? Put it in the comments. I'd love to know what you actually thought of that. And if you don't subscribe to the Pump Up Your Pod podcast, go and check it out. Go and subscribe to it. I'm not going to put even any hints on. I'm just going to drop an episode that's completely AI.

    [00:19:37] And if you then find it and listen to it, let me know what you think of that. Did it dupe you or not? If you like this kind of content, hit the like and subscribe button. I love. This channel, I'm really enjoying it, guys. I'm just loving doing it. So I'm going to continue to do it. And if you've got any ideas, feedback, things that you'd like me to do [00:20:00] topics on, go for it.

    [00:20:02] Put them in the comments. See you later.

 
 
Previous
Previous

When to say no to a brand deal

Next
Next

Microphone Battle: Shure SM7B vs Shure MV7