VO Infusion

Deb Munro & Keith Norton

AI is already in the booth, and most voice actors are getting their information about it from rumors, headlines and contract fine print. VO Infusion is where two working voice actors, Deb Munro and Keith Norton, sit down and sort out what’s real. We talk through the tools, the synthetic voice deals, the contracts and the protections that groups like NAVA and CAVA are pushing for, and we do it in plain English. Deb brings the energy and a little creative chaos, while Keith shows up with the questions and a healthy dose of skepticism. Together we’re trying to help voice actors understand AI well enough to make their own calls about it, without the doom and without the hype. New episodes come out in audio and video. Whether you’re booking your first jobs or you’ve been behind the mic for years, you’ll leave each episode knowing more than when you started.

  1. Oct 15

    Vocal Intellectual Property

    Are our vocal cords now intellectual property? A pretty big company asked me recently what a cloned version of my voice would sound like, so I built one. It took about an hour to get it somewhere I felt okay with, and that was the low bar. Then I recorded my own read of the same legal copy and put the two side by side. Mine ran a little quicker and had depth to it, and I told them they'd see the difference on the screen as much as hear it. Deb and I get into what happens when your voice becomes something a company wants to license and keep using after you leave. Getting a voice treated as intellectual property is hard, and smart, hungry people are working on it. So let's imagine we get everything we want... then what? Deb makes the case for jumping in, while giving the folks upset about AI-made artwork their due. What Deb and I get into - Turning a voice into property. Between NAVA's lobbying, Tennessee's ELVIS Act and a biometrics class action out of Chicago, there's movement. I think it comes down to the vocal print: establish that it's you, and you have standing. - Where the clone goes to work. Corporate narration may shift to clones the client directs on their own account. E-learning is the canary in the coal mine, since those courses need students to actually learn. - Two samples on one website. Deb floated posting a clone sample next to a real one, a clean way to show clients you get what you pay for. - Read the fine print. Deb is seeing "in perpetuity" in nearly every pay-to-play audition. A language model can flag trouble spots as a starting point, but on my big job my agent handles the contract, and if I have to pay him, I pay him. - New work from cheap tools. A small business can now make a commercial over a weekend. Microdramas have serious investor money behind them and could be where actors cut their teeth. - Freed up to be human. If the internet fills up with bots, there's an optimistic version where we get pushed back toward each other, the way young people have been drifting into trades like barbering. The line to take away Each of us is running a small, specific creative business, and that comes with a lot of opportunities to screw up. Start thinking like one now, and take Deb's advice: if you're afraid of this stuff, get in and find a way to control it. Mentioned in this episode - NAVA - The ELVIS Act (Tennessee) - ElevenLabs - Voices.com - Voice123 - ChatGPT - Claude - Netflix - Paramount VO Infusion is co-hosted by Deb Munro and Keith Norton. Find more at voinfusion.com.

    Vocal Intellectual Property
  2. Oct 15

    The Race to the Modal Average

    Why AI voices are starting to sound the same Deb's guitarist is, in her words, a dinosaur. He can't open an MP3 on his phone, and he's the one who told her AI music is going to make everything sound the same. He's dead on. Remember mean, median and mode from high school math? These tools scrub everything digitally available and race toward the most frequent answer, the modal average, because that's the version of credibility they're after. I see it on LinkedIn, which used to be a great place to build a business and is now a game of guessing who actually wrote the post. I spent a good while training my own model out of its tells, so you'll notice these notes contain no em dashes and no single-sentence paragraphs. What Deb and I get into - The tells. Mine include the em dash and the words "chaos" and "humanity." Deb's giveaway is emojis on every bullet. - The view from the decision maker's chair. Deb's music festival got 167 band submissions, and the generated ones put her off. Same goes for your note to an agent, so use the skeleton and then make it yours. - Train it how you write. I fed mine about 30 examples at each of three lengths, then kept correcting it until the habits stopped. Memory has a ceiling, so be picky about what it keeps. - Reads go in cycles. Deb traces the swing from hitting people over the head to the flat read and what she calls the Billie Eilish read. Now the broadcaster is making a comeback. - Bots talking to bots. I can picture a buyer agent sorting a thousand inbound pitches down to ten, and then it's up to a human to pitch me as a human. Sales might start to look like knocking on doors again. - Own your message. Generative engine optimization, the new SEO, leans hard on a brand's own material. I'm still confirming a stat on how much, but the move is clear: put your work and thoughts out there on a rhythm that suits you. J. Michael Collins is a good example of someone in our industry who does this well. The line to take away If everything is racing toward the modal average, make your version different. Creatives need to get into this conversation in more ways than putting our guard up, because we're all still learning how this stuff works and there's room to make it work for us. Mentioned in this episode - Pig Knuckle (pigknuckle.ai) - ChatGPT - LinkedIn - Google Flow - J. Michael Collins - Billie Eilish - Jack FM - Finding Your Own Work (Deb's class) VO Infusion is co-hosted by Deb Munro and Keith Norton. Find more at voinfusion.com.

    The Race to the Modal Average
  3. Oct 15

    Cloning - The Ghost in the Booth

    What happens when your voice leaves your body? Deb opened with news from an e-learning company she has voiced for over 20 years. Client demand for AI pushed them to cut their roster from 25 voices to five, and Deb is one of the five who stayed. The owner's answer is to clone voices and license them, so the actor gets paid for the original work and a royalty every time the clone is used, and that idea got both of us excited. I'm in favor of clones as a concept for a personal reason. Two people in my life died with ALS, and in their last years they couldn't tell their families they loved them, so whatever else we argue about, this technology can give a voice back. I also think anyone who hasn't cloned their own voice yet is missing out on understanding how these things actually work. What Deb and I get into - Where a clone helps. A client needed a one-line pickup months after a job, and my clone handled it. Deb adds that you can't always travel with your gear, since some countries will stop you at the border if they think you're working. - E-learning could split down the middle. Deb thinks about half of clients will stay with synthetic voices because of cost, while I think plenty of companies that never used a voice before will start hiring one. - AI as a scratch track. Deb uses it for her first pass on ideas and direction, then makes the work her own. - Faster safety copy. I timed the important safety information on a pharma spot: my own read ran about 37 seconds, and my clone plus a little speed in Adobe Audition got it to 26 while still honoring the pacing rules. That's more time for the brand to tell its story. - The fear is how it gets used. Tim Friedlander of NAVA told Deb the worry is that you can't control what a clone says once it's out there. NAVA is focused on that output side through the NO FAKES Act, Tennessee has the ELVIS Act, and voice actors and podcasters in Chicago are going after the input side under Illinois biometric privacy law. - How I built my clone. I recorded about two hours of myself just talking with family and friends, no script, and I turned off sharing so nobody else can use it. It fooled my mom and dad more than once, and now we have a code word for any phone call that involves money. - A licensing marketplace. I think it could work the way the Collegiate Licensing Company handles college logos, where you pay a fee, use the brand and the school gets its piece. Deb takes it further with a license that tells clients what level of actor they're getting. The line to take away I've been talking about licensing our voices since 2018, when I was told it was all covered in the buyout. It isn't, and the job now is to make clones something we own and license on our terms instead of something that gets used without us. Mentioned in this episode - Jesse Carroll - Mark Grau - Tim Friedlander - NAVA - NO FAKES Act - ELVIS Act (Tennessee) - ElevenLabs - Rime - Amazon - Adobe Audition - Bev Standing and TikTok - Collegiate Licensing Company (CLC) - Bodalgo - Voice123 - Voices.com - Pig Knuckle (pigknuckle.ai) VO Infusion is co-hosted by Deb Munro and Keith Norton. Find more at voinfusion.com.

    Cloning - The Ghost in the Booth
  4. Oct 15

    AI Agents

    Let the machines handle the boring parts Deb opened by asking what an AI agent even means for a voice actor, and I had to admit the word is overused, with probably five different definitions floating around. The version I care about is a system that works on your behalf, tying together LinkedIn, Instagram, email, texts, invoicing and editing so the repetitive grind gets done with one push of a button. Deb played guinea pig again. She's set up auto-booking for her coaching students but worries about looking unavailable to voiceover clients, and her house is full of tech she paid for that's still in the box. My answer is the same one I give everybody: before you automate anything, figure out what your business looks like, what you want it to look like and where the friction is. What Deb and I get into - Start with your own friction. Jesse Carroll has great systems, including file labeling for the 40 to 50 auditions he does a day, but they solve his problems, not mine. - Automate the mundane, keep the personal. Deb proofs and rewrites anything AI drafts for her, and I would never let an agent audition for me or answer an email on my behalf. - Chatbot versus agent. A chatbot answers your questions, while an agent actually takes action. I'd bet Deb's Sora refund came from exactly that kind of system, with no human involved. - Mistakes happen either way. Deb's automated festival post didn't go out one day, but she could just as easily have forgotten to post it herself. - The copy artist gets unlocked. Deb draws best when she can see a reference, and these tools give artists a base to make their own. If nobody's hiring you for a cartoon, make your own cartoon. - An idea that used to cost $50,000. Deb designed boot-shaped shot glasses for her band, Bootlegged, and a sponsor with a 3D printer is making them happen for little more than the cost of materials. - Wake up asking what's possible. I'd love a system that scours the bajillion auditions on Backstage and finds the one that's right for me. The line to take away Deb pulled this one from my essay: use AI to protect the time around your reads, not to replace the read. I've also been saying for almost a decade that if you let the nerds be nerds, they'll turn us into pets, so let's include ourselves in the conversation and decide what we want these tools to do for us. Mentioned in this episode - Jesse Carroll - LinkedIn - Instagram - Backstage - ChatGPT - Perplexity - Claude - Sora - Pig Knuckle (pigknuckle.ai) - Magic: The Gathering - Bootlegged (Deb's band) VO Infusion is co-hosted by Deb Munro and Keith Norton. Find more at voinfusion.com.

    AI Agents
  5. Oct 15

    The AI Dramaturg

    It's coming for your prep work When people used to ask what I did for a living, I'd tell them I did two things: I was a fundraiser at Stanford University and I was a voice actor. Nobody wanted to talk about Stanford. What they didn't see was the pile of work behind the voice part, from bookkeeping to troubleshooting software, and most of us carry it without an agent or a manager. That pile is what this episode is about. A dramaturg, in my shorthand, is the project manager of a theater production, making sure everything is buttoned up. That's how I think these tools earn their keep. I do understand people who want out of any conversation that smells like AI, but if you treat it all as a threat you never get to dabble. What Deb and I get into - Help with the minutiae. New to Adobe Audition? Skip the five minutes of YouTube preamble and get the one answer you need. It won't replace the ear of George the Tech. - Process nothing, learn the basics. Deb doesn't do any processing or noise removal, because once you process, the engineer can't undo it. Editing is another matter. De-breathing for long-form is on our list to learn, and if the tool you need doesn't exist, you can vibe code it with Claude. - The price of getting in. I think our industry has put up a moat that's expensive to cross, probably somewhere between $30,000 and $50,000 to build a career. Deb calls out demo pricing and I play the other side, since the market's the market. Monmouth University now offers voice acting courses, which is a start. - Who's coming for your job. The people who figure out how to use these tools are the ones coming for our jobs, which is a good reason to become one of them. Deb makes the case from her music festival promo: the idea is hers because she thought it through enough to ask for it. - Protection moves slowly. The Illinois voice biometrics case matters, and so does the NO FAKES Act that NAVA is pushing forward. We should own our data too, since on social media we've been the product the whole time. Deb's mentor Bob Bergen takes a felt marker to contracts before he signs. - Turn off model training. It's in your settings. My business partner and I won't use any model that doesn't let us turn training off, and we set up a trigger for terms-of-service changes. The line to take away Let these tools carry the prep and the paperwork so there's more of you left for the booth. Then go into your settings and make sure what you put into them stays yours. Mentioned in this episode - Stanford University - Adobe Audition - YouTube - George the Tech - Claude - ChatGPT - Monmouth University - NO FAKES Act - NAVA - Bob Bergen VO Infusion is co-hosted by Deb Munro and Keith Norton. Find more at voinfusion.com.

    The AI Dramaturg
  6. Oct 15

    Stop Chasing the Dashboard

    Why voice actors need to think like architects in the age of AI The question I get asked most is some version of "what are you using?" and it frustrates me, because I don't have a good answer. Usually I haven't tried the thing they're talking about, and they haven't heard of what I'm using. More keep showing up because regular people have figured out they can build them. Deb lays out the split plainly. A Chaser is always hunting for the next big platform, and an Architect asks how to work with it. We spend the episode on staying the second kind of person while the dashboards keep changing. What Deb and I get into - Builder culture. Things move this fast because people are building with the tools that already exist. The Global Voice Acting Academy's rate guide enhancement was built using Claude. - Solve your own problem first. When I dove in hard about a year ago, rule one was don't try to make money. Solve your problem and the rest takes care of itself. I ask whether I can build a tool instead of subscribing, though I happily keep paying friends whose tools I like. - Wait a week. If you don't like a tool, wait a week. If it's awesome, wait a week, because there might be a better one. Deb bought a year of Sora before it announced it was shutting down, and she wants to give them kudos for refunding her. - Patience in a fast market. Software now ships and updates at the same time. The lacrosse video game I finally got keeps getting fixed as people play it, so ask what you're trying to accomplish before you jump. - Let one tool steer another. I pointed Deb to Higgsfield, which makes you pick from a lot of video models. When you don't know which one, ask your other LLM and save yourself 25 takes of a person with five arms. - The door opens for the weirdo. Incubators used to bet on engineers and finance people. Now the kid who was our town's dungeon master is making a video game as a beta tester of Pig Knuckle, which Paul Langtry and I built. Deb gives the skeptics their due, and her answer is to put your own twist on it, like a cover band doing Zeppelin. The line to take away The tools will keep changing faster than any of us can track, and chasing every one of them wears you out. What holds up is knowing the problem you're trying to solve. Start there and let the dashboard catch up to you. Mentioned in this episode - Claude - Anthropic - Global Voice Acting Academy - Voice123 - Pig Knuckle - Alchemy Agentic - Paul Langtry - Sora - Higgsfield - Photoshop - OpenAI - Microsoft Copilot - Perplexity VO Infusion is co-hosted by Deb Munro and Keith Norton. Find more at voinfusion.com.

    Stop Chasing the Dashboard
  7. Oct 15

    Prompt Engineering, Part 2

    Prompt Engineering, Part 2: Am I replacing somebody? We picked up right where Part 1 left off, with the question Deb wouldn't let go of: if I use a persona prompt to get a workout plan, am I taking work from a personal trainer? My answer starts at the gym. Most people who pay for a membership go once in those first two weeks, probably to find out where the bathrooms are, so the trainer was never going to see them anyway. The same logic applies to our work. I could teach myself to make a demo at home and it would sound like crap next to one built with a real producer, and I'd still hire one for the demo that matters. Using these tools well helps you figure out who to hire and for what. What Deb and I get into - Make it cite its sources. If you're doing research, put that in the prompt, because it won't do it on its own. Even with sources built in, the one time Pig Knuckle got something wrong for me was because the article it pulled from was wrong. - Turn the tool around. Deb's take is that a trainer worried about losing clients can use the same tools to figure out how to keep them coming back. - Sell your expertise as prompt kits. Coaches can package what they know. After 25 years in fundraising I built 600 prompts for that world, and people who bought a few usually stopped once they understood how to build their own. - Clones hit a ceiling. I've listened to my clone side by side with my own voice, and there's no contest. My bet is that smaller companies leaning on clones will find the limits and start hiring humans, which means more voice work over time. - Plain language over jargon. Prompt engineering is articulating what you want with clarity, and Deb thinks it's a skill worth teaching the next generation. - Get into the conversation. A new lawsuit in Illinois over the biometrics of the voice is worth watching, because we either build our own or include ourselves. The line to take away There's nobody better than creatives to unleash what these tools can do and to make sure it gets done responsibly. That only happens if we're part of the conversation instead of standing outside it. Mentioned in this episode - ChatGPT - Google - Pig Knuckle (pigknuckle.ai) VO Infusion is co-hosted by Deb Munro and Keith Norton. Find more at voinfusion.com.

    Prompt Engineering, Part 2
  8. Oct 15

    Prompt Engineering, Part 1

    Prompt Engineering, Part 1: Direct it the way you direct yourself The prompt I lean on most these days runs about eight pages. It started as a paragraph, and it took months of telling it what I wanted, and just as often what I didn't want (em dashes, I'm looking at you), to get it there. Prompt engineering mostly comes down to getting what's in your head into the machine with clarity, and voice actors already do a version of that every day in the booth when we self-direct. This is the first half of a two-part conversation, with Deb playing guinea pig on purpose and asking the questions a lot of you are probably asking. Part 1 is about how to direct these tools. In Part 2 we get to the uncomfortable question of whether using them means replacing somebody. What Deb and I get into - Plain English is the new code. You used to need a programming language to make software do anything. Now you just say what you want, and when it misses, you get more specific. - Be over-descriptive. Deb compares it to writing, where the detail is the point. My analogy is that these tools are the smartest intern from MIT who doesn't know sh!t about sh!t, so give it structure and something to launch from. - Bring your self-direction into the prompt. Subtext, status, who you're talking to and what happened right beforehand all belong in there when you want ideas for an audition. - Two steps before. Asking for the moment before is useful. I ask for what happened two steps before, then find my own way from there through improv to the script, like inviting your buddy Mike over before the barbecue sauce spot. - Personas pull you off the average. Ask ChatGPT for three audition takes and you'll get something halfway decent. Tell it to act as a world-class condiment commercial producer and casting agent and you move it away from the most common answer, which Ben Affleck has called a race to the average. - Keep the human at the center. For an audition that matters, I want three versions: my own read, an AI-assisted read and a read directed by one of my coaches. The line to take away Strip away the jargon and prompt engineering is being able to articulate what you want and explain it clearly. Nobody does that better than people who direct themselves for a living. Mentioned in this episode - Pig Knuckle (pigknuckle.ai) - ChatGPT - Claude - Anthropic - Voice123 - Elaine Clark - Susan Blu, "Word of Mouth" - Ben Affleck and Matt Damon VO Infusion is co-hosted by Deb Munro and Keith Norton. Find more at voinfusion.com.

    Prompt Engineering, Part 1

About

AI is already in the booth, and most voice actors are getting their information about it from rumors, headlines and contract fine print. VO Infusion is where two working voice actors, Deb Munro and Keith Norton, sit down and sort out what’s real. We talk through the tools, the synthetic voice deals, the contracts and the protections that groups like NAVA and CAVA are pushing for, and we do it in plain English. Deb brings the energy and a little creative chaos, while Keith shows up with the questions and a healthy dose of skepticism. Together we’re trying to help voice actors understand AI well enough to make their own calls about it, without the doom and without the hype. New episodes come out in audio and video. Whether you’re booking your first jobs or you’ve been behind the mic for years, you’ll leave each episode knowing more than when you started.