Engineers Teach AI to Flirt and Confess Love

People are incredibly complex by nature, and voice plays a decisive role in helping us communicate with the world around us. On average, a person has more than 6,200 thoughts a day and more than 27 categories of emotions to help us understand what we feel. That is what makes speech so deeply dynamic.

What words come to mind when you think of artificial intelligence? Perhaps logic, coldness and lack of feeling? Against all odds, AI systems could become the new Romeo or Juliet.

The development team at Sonantic, a technology company innovating in AI voice platforms, has improved AI flirting by giving it the ability to breathe, sigh and laugh. These human-like sounds can add just a little of the emotional connection that any thriving romance requires.

"At Sonantic, we strive to imitate the human voice. Our AI-generated voice models can express a wide range of bold emotions. Yet so much verbal communication lies in subtleties. The best performances show true complexity, vulnerability and authenticity, and subtlety is a crucial component of skillful communication. As the old saying goes, 'it's not what you say, but how you say it.' No one understands this concept better than storytellers and actors. Our latest advances, subtle emotions and non-speech sounds, create unlimited potential for our clients like never before," the company said. 

Although these added natural sounds are new achievements, Sonantic can already speak playfully and teasingly, which matters for any aspiring Romeo or Juliet.

To get a closer look at this impressive development, check out Sonantic's promotional video on YouTube. It's Valentine's Day. The girl's voice sounds completely natural and very soothing, especially with the sighs and laughter so beautifully woven into her speech.

That's why it is shocking when the girl suddenly makes her revelation: "I'm not real. I was never born. And I will never die. Because I do not exist."

Sonantic also demonstrated the interface of the program used to generate the voice. In it, the user can change the speed of speech, adjust intonation, style and tempo, and add various sounds between words.

Perfecting subtle emotions is more difficult from a technological standpoint. AI voice models tend to make emotions sound muffled. The more subtle they are, the greater the chance they get drowned out and sound robotic. Engineers explored methods to ensure their voice models counteract this muffling effect and keep emotional subtlety intact. The company introduced new styles for clients, such as "flirty," "shy" and "teasing."

When you think about it, communication is much more than words. There are pauses, deep breaths, moans and gasps that give extra insight into what we think and feel at any moment. It is these non-speech sounds that bring our conversations to life. Even something as simple as breathing, which seems ordinary to us, affects how a natural human voice sounds. Without these small details, our conversations would sound mechanical and monotonous.

Sonantic reported that its AI voice models can now breathe, laugh and even mock. The company has always worked with the most talented voice actors to train its voice engine, and this latest evolution was no different. But subtle emotions can be challenging even for the most experienced performers, and it often takes several takes to convey them correctly. 

Engineers reworked scripts, spent more time on preparation, recorded more takes and gave actors maximum creative freedom. Ultimately, this produced dynamic, versatile and multilayered material to work with.

Capturing non-speech sounds was not easy, and researchers had to experiment with new algorithms until they achieved the desired result. In the first few attempts, everything sounded distorted and robotic. But thanks to excellent collaboration between the research, information and engineering teams, they developed new patented algorithms that made it possible to capture these subtleties.

To showcase their innovative achievements, the engineers made a video titled "What's Her Secret?" This is the most realistic product demonstration Sonantic has ever created.

"When we came up with the concept for the video, we decided to focus on the theme of love, because that is when we feel most vulnerable. The calm, soothing voice of the girl wavers slightly as she asks, 'What would it take for you to fall in love with me?' As you listen to the monologue, notice when she laughs, breathes or sighs at just the right moments. The video surprises viewers when it turns out that the woman on screen is a real person, while the speaking main character is artificial intelligence," the company said. 

These advances in speech synthesis make the Sonantic platform more complete. The company is pleased to offer its clients new capabilities. Engineers said they are building truly powerful tools for voice actors and entertainment studios to keep advancing voice innovation in the future.

The company's voice synthesis is already used in the production of Hollywood films and video games. It can express joy and sadness, and now flirtation too. In August 2021, Sonantic applied its technology to recreate the voice of actor Val Kilmer, who lost the ability to speak as a result of treatment for throat cancer.