Latest / The DTC Marketing Podcast with Fexingo: Direct-to-Consumer Brands, Shopify, and E-Commerce Growth / How DTC Brands Are Using AI Voice Cloning for Brand Audio
Transcript
- Lucas: So a few weeks ago I heard a radio ad for a direct to consumer mattress brand — sounded exactly like the founder, that same slightly raspy voice he uses in all his video content. Only problem is, I happen to know that founder is currently on paternity leave and hasn't recorded anything in months. Luna: Wait — was that an AI clone? I've been seeing more brands do this, but I didn't realize it had reached radio. Lucas: Yeah, it was an AI voice clone. And it got me thinking: this is probably the fastest-growing DTC marketing tool nobody's talking about yet. I'd say at least a dozen brands I follow have started using synthetic versions of their founder's voice for ads, onboarding sequences, even abandoned-cart audio reminders. Luna: Let's back up — how does this actually work? Are we talking about a few seconds of the founder talking and then the AI generates the rest? Lucas: Exactly that. You give a voice-cloning platform — there are a handful now, like ElevenLabs and Respeecher — about ten to fifteen minutes of clean audio of the person speaking. The model learns the cadence, the pitch, the little verbal tics. Then you can type any script and it reads it in that voice. The newer models even capture emotional inflection. Luna: So a brand that used to book a studio session for every new ad — now they just type out the copy and hit generate. Lucas: Right. And it's not just cost savings — it's speed. I talked to a supplement DTC brand that used to spend three weeks per ad campaign just on voiceover production: booking talent, recording, retakes. They cloned the founder's voice last quarter. Now they turn around an ad in two days. Their cost per finished minute of audio dropped from about eight hundred dollars to basically zero. Luna: Zero is a compelling number. But I have to ask — does the listener know? I think there's a trust question here. Lucas: That's the big one. Right now, disclosure is voluntary. The FTC hasn't issued specific rules for synthetic voice in ads yet, though they've flagged it in their recent comment period on AI. Some brands put a tiny 'ai generated voice' in the fine print. Most don't. And the supplement brand I mentioned — they don't disclose at all. Luna: That feels risky. If a loyal customer finds out later that the founder never actually recorded those messages, doesn't that erode trust? Lucas: It could. But here's the counterpoint: the founder is saying the same things he would say. The clone isn't making false claims — it's delivering approved copy. The brand's argument is that the voice itself is a brand asset, like a logo. You don't expect the CEO to personally draw every logo placement. Luna: I see the analogy, but a logo isn't a person's actual voice. A voice carries implied authenticity. If I hear the founder talking about how much he loves the product, I assume he's the one talking. Lucas: And that assumption is exactly what makes the tool so effective. I think the smart brands are going to land on a middle ground: use the clone for volume — onboarding sequences, retargeting audio, A-B test scripts — but have the real founder record the hero ads. Or at least put a clear disclosure that doesn't hide in the footer. Luna: There's also the creative upside. If you can iterate on audio copy as fast as you iterate on email subject lines, that changes the game for testing. Lucas: Massively. Before voice cloning, testing five different ad scripts meant five studio sessions. Now you type five versions, generate them in fifteen minutes, run them against each other on Spotify or podcast spots, and the winner goes live by end of day. That's a structural advantage for DTC brands that move fast. Luna: I want to talk about personalization at scale, because that's where this gets really interesting to me. Imagine an audio ad that addresses the listener by name and references their specific browsing behavior. Lucas: That is happening right now. There's a DTC skincare brand — I won't name them because they're still piloting — that's using a cloned voice to generate dynamic audio messages for their SMS flows. If you abandoned a cart with a moisturizer, you get a thirty-second audio clip from 'the founder' talking about why that moisturizer is perfect for dry skin. It's personalized, it's contextual, and it sounds human. Luna: And the founder doesn't have to record a thousand variations. They record one set of base messages, and the AI stitches in the product name and the customer's first name. Lucas: Exactly. The early data from that pilot shows a forty percent higher click-through rate on the audio messages compared to standard SMS text. Which makes sense — hearing a voice is more engaging than reading a block of text. Luna: So where do you see this going in the next year? Is every DTC brand going to have a voice clone? Lucas: I think many will. The barrier to entry is dropping fast. You can get a decent clone for under a hundred dollars a month now. But the differentiation will come from how brands handle the authenticity piece. The ones that are transparent about it — 'hey, this is an AI version of me, but I stand behind every word' — those will build trust. The ones that try to deceive will get caught, and it will backfire. Luna: There's also the risk of misuse. What if someone clones a founder's voice without permission? We've already seen deepfake audio scams in other contexts. Lucas: That's a real concern. Most cloning platforms now require a live verification — you have to read a specific phrase while recording to prove consent. But it's not foolproof. Brands need to treat their voice data as carefully as they treat their customer data. Limit access, monitor usage, have a takedown process. Luna: It feels like we're at the same point with voice cloning that we were with ai generated product photography about a year ago. Everyone's excited, a few early adopters are seeing big wins, and the ethical norms haven't settled yet. Lucas: That's a great comparison. And look, I think this technology is genuinely useful for DTC brands. It lowers the cost of audio production, enables personalization, speeds up testing. But it also puts a new responsibility on marketers to be honest with their audience. Luna: Speaking of honesty — and I know this is a bit of a pivot — but one thing I really appreciate about this podcast is that we don't have to dance around ad reads or sponsored segments. We can just talk about what's actually working. Lucas: That's by design. We deliberately don't run ads on these episodes, and we've heard from a lot of listeners that they value that. If you want to support that choice, the link is buy me a coffee dot com slash fexingo. No pressure — just an option if the show has given you something useful. Luna: Yeah, it's a small gesture that keeps the conversation clean. And honestly, it helps us keep doing deep dives like this one. Lucas: Alright, back to voice cloning. One more thing I want to highlight: the international angle. If you're a DTC brand selling in five countries, you can clone the founder's voice and have it speak fluent Spanish, French, or Japanese — with the same emotional tone. That used to require hiring five different voice actors. Luna: Is the quality there for non-English languages? I've heard some synthetic voices in Spanish that sound robotic. Lucas: It depends on the platform and the language. English is the most mature. Spanish and French are getting good. Japanese and Korean are improving fast. The key is training the model with enough native-language data. Some brands are actually cloning a native speaker for each market rather than cloning the founder in multiple languages. Luna: That's a smart workaround. Use the founder's voice for brand recognition but a local voice for cultural authenticity. Lucas: Exactly. And that's where the real art comes in — knowing when to use the clone and when human authenticity matters more. I think the winning DTC brands in 2027 will be the ones that treat voice cloning as a tool in the toolbox, not a replacement for genuine human connection. Luna: I'd love to see some A-B test results on disclosure. Like, does a clear 'ai generated voice' label hurt conversion? Or does it actually build trust? Lucas: I haven't seen a rigorous study yet, but anecdotally, the brands that disclose are seeing no drop in engagement. Some listeners even comment that they appreciate the honesty. So I suspect the fear of disclosure is overblown. Luna: That matches what we've seen in influencer marketing — audiences can tell when something is fake, and they reward transparency. Lucas: Alright, I think that's a solid place to wrap. Voice cloning is coming fast to DTC marketing. The brands that use it thoughtfully, with disclosure and a clear strategy, are going to have a real edge. The ones that hide it are playing with fire. Luna: And we'll keep an eye on the regulation side. Thanks for the conversation, Lucas.