Highlights:

  • WaveForms AI’s audio LLMs will detect emotional nuances in human voices, interpret their meaning, and respond with appropriate emotions.
  • WaveForms AI focuses on AI voices that understand and respond to emotional cues, aiming to create AI experiences distinct from traditional text-based interactions like ChatGPT.

A startup aiming to revolutionize audio-based AI interactions, WaveForms AI, has secured USD 40 million in seed funding.

The recent funding round, led by Andreessen Horowitz, values WaveForms AI at an impressive USD 200 million from the outset. This investment not only reflects strong confidence in the startup but also highlights the increasing interest in AI voice technology and its potential to enable more realistic human-to-machine interactions.

WaveForms AI, headed by former OpenAI researcher Alexi Conneau, distinguishes itself by specializing in AI voices capable of interpreting and responding to emotional cues. The company seeks to redefine AI interactions, moving beyond the conventional text-based exchanges seen with ChatGPT and similar chatbots.

Conneau shared with Fortune that his goal is to create more natural and seamless interactions between people and AI by developing technology that emulates humanlike conversations. With his extensive experience, including leading the development of ChatGPT’s voice capabilities at OpenAI, he is well-equipped to achieve this vision.

According to Conneau, WaveForms AI’s dedicated focus on audio large language models will enable the creation of AI bots capable of understanding the nuances of human emotions in conversations. Unlike traditional chatbots that rely on text inputs, WaveForms AI aims to develop digital assistants that interact with users in a more intuitive, natural, and dynamic manner. The goal is to enhance AI communication, allowing chatbots to respond intelligently to the emotional context of their interactions with people.

This technology holds immense potential, with far-reaching implications. It could enable chatbots to deliver more lifelike and emotionally intelligent responses, enhancing user engagement and satisfaction. By fostering personalized and empathetic interactions, AI systems could better interpret emotional cues and respond more appropriately. If Conneau’s vision succeeds, it might reshape consumer expectations for AI, driving demand for applications with more humanlike qualities.

Conneau revealed to Fortune that the inspiration behind WaveForms AI’s technology comes from the film Her, which portrays a man’s deepening emotional connection with an AI assistant named Samantha. “When they see the technology, they think about the movie right away,” Conneau said, referencing the company’s AI voices.

Conneau’s work at OpenAI sparked controversy when one of the ChatGPT voices he developed closely resembled actress Scarlett Johansson, who portrayed Samantha in Her. Johansson later disclosed that she had been approached to license her voice for the project but declined. She eventually threatened legal action against OpenAI over the matter.

With WaveForms AI, Conneau aims to push the limits of audio AI, describing it as the “social-emotional layer” of artificial general intelligence—a vision of AI capable of performing tasks on par with humans. His goal is to improve AI’s ability to understand and respond to human language, enabling it to recognize nuances like tone, inflection, and accents.

WaveForms AI’s audio LLMs will be designed to detect the emotional nuances in human voices, interpret their meaning within the context of a conversation, and respond with emotionally expressive replies.

Conneau envisions intriguing applications for this technology, such as an audio LLM for teaching that can sense when a student is frustrated and respond with a more patient and supportive tone.

Conneau revealed that the startup is working on a “consumer software product” centered on fostering emotional connections between users and AI systems, though he did not share additional details.

“The idea is to create new, more immersive experiences with AI, ones that feel more enjoyable. There are many companies right now that are focusing on superintelligence, but not so much on the quality of the human-computer interaction,” he said in another interview.