Candy Voice: the A.I. movement where fake voice is now a reality available from a French company on your iPhone. What does this mean for news and truth? Scott Goodson finds out more today on the Uprising Pod. For more ideas on Uprising and movements, cultural movements and movement marketing, follow Uprising!!! on Facebook. We’ll continue to publish brand-new columns on a regular basis. Hey, do us a favor and please give Uprising!!! a review on iTunes. Scott Goodson is the author of best-selling book ‘Uprising: how to build a brand and change the world by sparking cultural movements,’ available on Amazon.com. Scott has helped create and build some of the world’s most iconic brands. He is founder of StrawberryFrog the world’s first movement marketing agency.
Uprising Interview transcript
SCOTT: You’re ahead of this technological movement in that your software can create these voice clones in record time. You can create voice clones quickly.
JEAN LUC: Yes, very quickly. For the smartphone application, when the user records 160 sentences, just 45 seconds after, the model is ready to produce his voice.
INTRO: Welcome to Uprising. Each episode looks inside what it takes to make the most dynamic and successful and cultural movements. Some of them in the business world, some in the social realm, some in politics, and some in between to see why people start uprisings. What gives those initiatives momentum and keeps them going and most important what lessons can you learn from these movements and how to apply them to your business and even personal life. Let’s explore the secret to sparking movements that move people into action.
Passionate ideas, controversial ideas, uprising ideas. The power is now in the hands of anyone to start a cultural movement.
SCOTT: Today we look at the artificial intelligence movement, something that has the domains of Silicon Valley, proclaiming with great fear in their voice: beware. Some anticipate an extraordinary future. And some–like Elon Musk–have warned us about the approaching apocalypse. Donald Trump fights for coal jobs, while Democrats that they great threats to jobs in America is artificial intelligence and automation. In a recent article in the New York Times, it was announced that a large percentage of investment jobs on Wall Street have already begun falling away to AI.
In the 2014 film Ex Machina, a billionaire invites a young programmer to spend a week in his remote estate to participate in a test to determine if the artificial intelligent female robot he created can pass as a human. At first, it seems like nothing–just a docile machine. Eventually, we learn the robot understands the depths of human nature and possesses all human emotion. This is the gray zone. This is where technology meets human beings. This is where Siri evolves. This is the rise of the robot movement or the AI movement. It forces us to ask what it means to be human and how technology plays a role in our lives. More importantly, should these robots be assistants that help us with menial tasks? Or should they be smarter us, do everything you can and more?
Here to help us answer these questions and explore this technological movement is Candy Voice’s creator Jean-Luc Crebouw direct from Paris, where he is speaking to us in the sweltering heat of Paris. Jean-Luc, welcome to our show today.
JEAN-LUC: Thank you for inviting me to your show.
SCOTT: Maybe a starting point is to understand what Candy Voice is and what Candy Voice can do.
JEAN-LUC: Candy Voice is a company born two years ago. We are specialized in voice processing. We sell eight different voice software and voice processing, for example, we use no filter. Today is voice conversation. More than 10 years, I’ve worked on voice processing. I realized voice conversion software to produce the voice of any body.
SCOTT: In effect, what Candy Voice is doing is cloning your voice. The result is that your voice can be cloned and used in technology. Now, I think Candy Voice sounds a little bit synthetic but in the future, will you be able to tell the difference between a human voice and your voice on Candy Voice?
JEAN-LUC: Little bit synthetic. On our website, for example, we have a demonstration but our technology because we need so little in our world to listen, to produce a voice. Only a few sentences are needed. We don’t use technology, for example, that uses 10,000 sentences to learn. That’s why we use synthetic voice.
SCOTT: Understood. About how many sentences do you need to record on your phone?
JEAN-LUC: You speak about the phone. In fact, our technology is used by professional and individual. Professional, we need 260 sentences for an actual voice. For an expressive voice, we use also 260 sentences. But for individual and our application on a smartphone, we only use 160 sentences. In fact, we couldn’t use less sentences. Sometimes the recording is not very good, so we kept only good sentences.
SCOTT: Have you tried to clone the voice of Donald Trump?
JEAN-JUC: No. Not yet.
SCOTT: When you record voice, can you pick up dialects and moods and reflections in the voice?
JEAN-LUC: Our technology is real time voice conversation. For example, you could hear me–not with my voice–in real time but a reimagined voice. That’s why we can’t reproduce the ?? In fact, when we use a text to speech or synthesize a former text, we can reproduce half of the quality.
SCOTT: How does your text to speech software work?
JEAN-LUC: We don’t have a text to speech software. We have a voice conversion software. We can customize any text to speech in the world. Do you see the difference?
SCOTT: I do see the difference. Yes.
JEAN-LUC: If we use genuinely voice of text to speech, we can customize your voice of the text to speech and have a million of voice with only one voice.
SCOTT: I can envision the usage of voice cloning for people who are going to lose their voice. Patients for example that have throat cancer or other types of ailments that lose their voice instead of having a voice that’s generated from a computer, they can keep their own voice. Do you see a risk for this application to be used, for example, for employees talking to their bosses or husbands and their wives? Do you see the conflict?
JEAN-LUC: We could, but we will not give it to anyone the possibility to do that. This particular we are using is dubbing for example.
JEAN-LUC: Yeah, dubbing.
SCOTT: In other words, a film that was produced in France could easily be cloned into English.
SCOTT: Ok. That’s interesting.
JEAN-LUC: Our proposal is to produce a voice for the user with his writing. I write something and anybody who has my own voice model could listen to me with my writing.
SCOTT: Do you think that it will be possible in the future for you to hear your voice on a recording and not know the difference whether it’s you or a clone.
JEAN-LUC: It’s possible, yes.
SCOTT: About how far in the future will that be?
JEAN-LUC: For example, our market is a robot. We want to give a robot a familiar voice. In this case, to have the best voice prediction we need the best voice recording. We are a smart phone application. If we have a good recording, we could have a good reproduction. The best way for people who really love their voice, it is preferable in English.
SCOTT: And that is the highest quality reproduction.
SCOTT: When you started Candy Voice, what was your goal at the beginning? What started you on the pathway to cloning voices?
JEAN-LUC: The goal started very, very, very early. When I was 4-years-old, I had an accident. I saw that I am a neuro and saw that I can stop a car running the road very fast. I stopped the car. I was a neuro. But it was more complicated and more difficult. When I touched the ground, my tongue was outside my mouth. My mouth was not completed fully. It was hard to speak again. I know that our voice is very important. Because I lost my voice, I tried to understand how the voice works. I found it. I found Candy Voice to reproduce Candy Voice. I have 60, 70 years. 53 years ago. Now, I can realize my dream.
SCOTT: Your personal experience of not being able to speak has been your catalyst that made you start thinking that this is an important thing to have. As you started out, you were thinking primarily of helping people with health-related issues. Was that the genesis of this?
JEAN-LUC: It was one of my goals. Yes, yes.
SCOTT: As a young child, did you want to sound like Darth Vader?
SCOTT: Or a famous singer?
JEAN-LUC: Before he was born.
SCOTT: Maybe you could be a famous karaoke singer. You know you borrow someone else’s voice.
JEAN-LUC: In fact, with voice transformation, I win a prize. [Speaks French]
SCOTT: The grand prize of the innovation competition in Geneva. How old were you?
SCOTT: That’s wonderful.
SCOTT: WE are living in a time if you have Google home or Siri on your iPhone or if you have an Amazon Echo–Alexa as they call it–we are living in a time when a personal, robotic voice is quite common. In fact, we probably don’t think about it but how voices are very much a part of our lives already. Candy Voice at first sounds like oh my gosh this is scary. But, on second thought you go hang on a second. We already around us have a lot of synthetic voices. Back in the 1970s, my father bought a Nissan car–his first Japanese car he bought in the 1970s. It was called a Nissan Maxima. One of the reasons he loved the car was because it had a female voice that talked to him. He would always talk about how much fun he had had a positive female voice telling him to turn off the car and to turn off the lights. I think when I was driving it, the voice would say, you better be careful otherwise you’ll be in trouble when your father gets home. Voice has been around for a long time. But the scary thing now, for me anyway, is what is the risk of hearing a voice on your phone that is someone who you know or respect or a public figure, telling you something that you just don’t know whether it’s true or it’s false. Have you thought about that?
JEAN-LUC: Yes. You know that it’s false. You understand?
SCOTT: How will you know?
JEAN-LUC: If somebody writes something to you and you listen to the voice, you know the original writer is a good person. If I call you with a telephone and the reproduced voice, sure you will have some trouble.
SCOTT: There are checks and balances to manage this process. We know that it’s not a computer talking and know that it is a credible human.
JEAN-LUC: For me, our goal is to personalize the product with any voice you want.
SCOTT: Go ahead.
JEAN-LUC: Sometimes when I make a voice conversion, I make a mistake. It is me, but it is not me. That is why we can have trouble.
SCOTT: There is a way of verifying the voice to make sure that it’s an authentic voice rather than a fake voice.
SCOTT: Understood. Why do you think there is a need for technology to talk to us?
JEAN-LUC: Thanks to voice recognition, five years ago voice recognition was not very good. Now, voice recognition gives you five percent error. Now we can find a lot of applications and understand what you say–not only for voice recognition but also for artificial intelligence. We can have a better recognition, so we need a better synthesize.
SCOTT: Do you find that there are clients today working with you–are you actually working with companies and organization to use this technology to improve their business?
JEAN-LUC: We have some prospects and customers in entertainment, in dubbing, in robotics and assistant robots. The voice would be given to a robot–a familiar voice. An old man or old woman can have the voice of their dead spouse. [French]
SCOTT: Oh, Alzheimer’s disease. They have deteriorating faculties and so they can’t speak so the voice can be used to
JEAN-LUC: The voice is very important for us because we recognize the voice. Our robots can help this person to have a voice that they can recognize.
SCOTT: By 2020, you’re sitting on a beautiful boat, enjoying your life and you’re very successful. What will you have achieved in four years? Where will voice cloning be in four years time?
JEAN-LUC: I hope lots of things. My goal: I want to start with a person who had a lot lesser voice. Everybody lost their voice. Children lost their voice. All children lose their voice. Old people between 20 and 60 lose their voice. An old person loses their voice. We want to give the possibility to everyone a memory of their voice. When I speak in a microphone and listen to me with a voice of a child. He had a great emotion. We want to welcome this. I think this is the future.
SCOTT: What type of song do you like to sing in karaoke?
JEAN-LUC: I’m not a singer.
SCOTT: Can you sing?
JEAN-LUC: Barry White. [laughs] The recording was not very good. We recognize it. I forgot to tell you that when I create a voice conversion model, I can speak in every language. For example, if you go on our website, we give the ability to speak in any voice in French voice. We forgot to tell people it’s only in French. So, we saw people writing English.
SCOTT: In translation.
JEAN-LUC: Let’s see if you can follow me. This is a Barry White song. Do you recognize it? Can you sing with me on this?
SCOTT: [singing]My darling. I can’t get enough of your love babe.
Do you know that song? It’s an old Barry White song.
JEAN-LUC: I could give you his voice if you want.
SCOTT: Can you do that? We will do a follow-up interview. Maybe if you can do a cloning of the Barry White song “Can’t Get Enough of Your Love Babe.” That would be a good one. Why do you call it Candy Voice?
JEAN-LUC: Because it is very difficult to find a website name with voice. It’s very difficult.
SCOTT: So you came up with Candy Voice. It’s a great name.
JEAN-LUC: Yes, I love it. Candy is for individual. Voice is for professional. That’s why I find this thing.
SCOTT: That’s why I call my company StrawberryFrog. It’s the same reason. But, I think if you lick Candy Voice, it’s sweeter than if you lick StrawberryFrog. So, how does the movement of artificial intelligence, how does Candy Voice fit into this movement? How important is it to develop artificial intelligence that is helpful?
JEAN-LUC: With a familiar voice, it is more friendly to have a voice that you know instead of a generic voice. We are not in artificial intelligence. But, we propose to artificial intelligence to give the possibility to old voice.
SCOTT: Candy Voice is cloning of voices: scary, fascinating, exciting. If I want to find out about it, if I am listening to the show and want to find out more about Candy Voice, where do I go? Do you have a website? Can I download an app?
JEAN-LUC: Yes. Our website, candyvoice.com. We have a smartphone application for an individual. It’s called Candy Voice Recorder. It is not completely finished for French today. For English, you have to wait two weeks. We start the beginning of February, but we had a lot of technical problems.
SCOTT: Thank you so much JC. This has been a fascinating discussion. Next time, I look forward to testing our Candy Voice so that I can sing in English and maybe we can hear it come across with the famous French singer, like Johnny Holiday maybe?
JEAN-LUC: It’s possible.
SCOTT: Or maybe you can sing in French and we can hear Katy Perry. Thank you so much for spending time with us on the Uprising podcast. Really fascinating. Merci.