Former OpenAI Lead Launches WaveForms AI: A $200M Venture into Empathetic Audio AI

Former OpenAI Lead Launches WaveForms AI to Tackle Speech Turing Test

Former OpenAI researcher Alexis Conneau has launched WaveForms AI, a new startup focused on improving spoken conversations between humans and machines. "I'm excited to announce the creation of WaveForms AI - an Audio LLM company that aims to solve the Speech Turing Test and bring Emotional Intelligence to AI," he said in a post on X. Conneau is one of the world's leading audio and text LLM researchers and co-creator of OpenAI's GPT-4 Advanced Voice Mode. He left the company in August and began working on WaveForms in September. The San Francisco-based startup aims to create AI audio software that can detect emotional cues and enable more natural verbal interactions with computers. "We're looking to make conversations so natural that users can't differentiate between speaking to a human or an AI," Conneau said. The company secured $40 million in seed funding from Andreessen Horowitz, valuing it at $200 million. "EQ is equally, if not more, important than IQ. To create the most immersive and human-like experiences, we need AI that truly feels like talking to a person. This includes giving AI a voice," said a16z in a statement. WaveForms AI currently operates with a lean team of just five employees, including Conneau and his co-founder Coralie Lemaitre, who previously worked in product strategy at Google. Conneau acknowledged the challenges ahead, saying, "We're aware of the risks, including the possibility of users becoming too attached to AI characters. It's something we're mindful of as we develop our models." The company has yet to release any public demos or product details, with Conneau revealing that WaveForms' first products will launch in 2025. "It's too soon to dive into details, but expect us to prove our technology in the consumer space before expanding into business-to-business applications," he said.

Former OpenAI researcher raises $40 million to build more empathetic audio AI

SAN FRANCISCO (Reuters) - - A former OpenAI researcher has raised $40 million for a startup that will build artificial-intelligence models that connect emotionally with humans via voice, the company said on Monday. The startup, WaveForms AI, is backed by Andreessen Horowitz and valued at $200 million, CEO Alexis Conneau told Reuters. Conneau co-created the voice mode capability on OpenAI's GPT-4o model, which when released earlier this year showed off the capability to respond in real time with no delay, as well as handle interruptions - both hallmarks of realistic conversations that had to date eluded AI voice assistants. After the launch, OpenAI CEO Sam Altman posted the word "her" on social media platform X, in what appeared to be a reference to the 2013 film of the same name by Spike Jonze about a man who fell in love with his AI assistant, voiced by Scarlett Johansson. OpenAI's launch of GPT-4o became controversial when Johansson accused the startup of copying her voice in the movie. OpenAI said the voice was not an imitation of Johansson and that the company used a different voice actress. Conneau, along with WaveForms AI cofounder Coralie Lemaitre, said the startup will use the funding to train new audio AI models that solve the problem of making voice conversation with an AI bot feel indistinguishable from a human. He added that audio is the key to making AI interactions feel "deeply human." "Audio ... conveys emotions and provides emotional responses back to users," he said. Conneau said the company plans to build a consumer software product that will focus on the emotional connection between the user and the AI, but declined to provide more details. "The idea is to create new, more immersive experiences with AI, ones that feel more enjoyable," he told Reuters. "There are many companies right now that are focusing on superintelligence, but not so much on the quality of the human-computer interaction." (Reporting by Anna Tong in San Francisco; Editing by Matthew Lewis)

WaveForms AI raises $40M in seed funding to create more empathetic AI voices - SiliconANGLE

WaveForms AI raises $40M in seed funding to create more empathetic AI voices A startup called WaveForms AI is looking to transform audio-based artificial intelligence interactions after closing on a bumper $40 million seed funding round. Today's round was led by Andreessen Horowitz and values WaveForms AI at a staggering $200 million right off the bat. The investment doesn't just underline the confidence in the startup, but also illustrates the growing interest in AI voice technology and the potential for more lifelike human-to-machine interactions. WaveForms AI is led by former OpenAI researcher Alexi Conneau, and sets itself apart from other startups by focusing exclusively on AI voices that can understand and respond to emotional cues. It's aiming to create AI experiences very different from the traditional text-based interactions people have with ChatGPT and other chatbots. Conneau told Fortune that he's looking to create more seamless and natural experiences for people when they interact with AI, and to that end his company is focused on creating technology that can mimic human-like conversations. He has the skills and experience to do this, having previously led the development of ChatGPT's voice capabilities while working at OpenAI. WaveForms AI's laser focus on audio large language models will lead to the development of AI bots that can interpret the intricacies of human emotions during conversations, Conneau said. Whereas traditional chatbots are reliant on text inputs, WaveForms AI wants to create digital assistants that can engage with their users in a more intuitive, natural and dynamic way. It aims to make communication with AI more sophisticated, so chatbots can react to the emotional context of conversations they have with people. There are vast and varied potential implications for this kind of technology. It can result in more lifelike and emotionally intelligent responses from chatbots, increasing user engagement and satisfaction. It could also pave the way for more personalized and empathetic interactions, enabling AI systems to pick up on emotional cues and respond in a more appropriate way. If it works as Conneau hopes, it may soon transform consumer expectations around AI, increasing the demand for more human-like qualities in AI-powered applications. Conneau told Fortune that the inspiration for WaveForms AI's technology comes from the movie 'Her', which tells the story of a man's increasingly passionate relationship with an AI assistant called Samantha. "When they see the technology, they think about the movie right away," Conneau said of the company's AI voices. Conneau's work at OpenAI attracted some controversy, as one of the voices he created for ChatGPT bore a striking resemblance to the actress Scarlett Johansson, who played the role of Samantha in Her. Johansson later revealed she had been asked to license her voice for the project, but refused to do so, and later threatened to take legal action against OpenAI. With WaveForms AI, Conneau said he wants to push the boundaries of what audio AI can be, saying that he sees it as the "social-emotional layer" of artificial general intelligence, which is a vision of a future AI system that can do certain things as well as humans. His goal, he said, is to enhance the ability of AI to understand and respond to human language, so it can interpret details like tone, inflection and accents. For instance, WaveForms AI's audio LLMs will be able to capture the emotional subtleties of human voices, understand what they imply in the context of the conversation, and then reply while conveying their own emotions in return. This could result in some interesting applications, Conneau believes, such as an audio LLM for teaching that's able to detect when a student is frustrated, and respond by adopting a more patient approach. According to Conneau, the startup is planning to release a "consumer software product" that's focused on the emotional connection between users and AI systems, but he didn't provide any more details than this. "The idea is to create new, more immersive experiences with AI, ones that feel more enjoyable," he told Reuters in a another interview. "There are many companies right now that are focusing on superintelligence, but not so much on the quality of the human-computer interaction."

Former OpenAI researcher raises $40 million to build more empathetic audio AI

SAN FRANCISCO, Dec 9 (Reuters) - - A former OpenAI researcher has raised $40 million for a startup that will build artificial-intelligence models that connect emotionally with humans via voice, the company said on Monday. The startup, WaveForms AI, is backed by Andreessen Horowitz and valued at $200 million, CEO Alexis Conneau told Reuters. Conneau co-created the voice mode capability on OpenAI's GPT-4o model, which when released earlier this year showed off the capability to respond in real time with no delay, as well as handle interruptions - both hallmarks of realistic conversations that had to date eluded AI voice assistants. After the launch, OpenAI CEO Sam Altman posted the word "her" on social media platform X, in what appeared to be a reference to the 2013 film of the same name by Spike Jonze about a man who fell in love with his AI assistant, voiced by Scarlett Johansson. OpenAI's launch of GPT-4o became controversial when Johansson accused the startup of copying her voice in the movie. OpenAI said the voice was not an imitation of Johansson and that the company used a different voice actress. Conneau, along with WaveForms AI cofounder Coralie Lemaitre, said the startup will use the funding to train new audio AI models that solve the problem of making voice conversation with an AI bot feel indistinguishable from a human. He added that audio is the key to making AI interactions feel "deeply human." "Audio ... conveys emotions and provides emotional responses back to users," he said. Conneau said the company plans to build a consumer software product that will focus on the emotional connection between the user and the AI, but declined to provide more details. "The idea is to create new, more immersive experiences with AI, ones that feel more enjoyable," he told Reuters. "There are many companies right now that are focusing on superintelligence, but not so much on the quality of the human-computer interaction." Reporting by Anna Tong in San Francisco Editing by Matthew Lewis Our Standards: The Thomson Reuters Trust Principles., opens new tab Suggested Topics:DisruptedDisrupted Anna Tong Thomson Reuters Anna Tong is a correspondent for Reuters based in San Francisco, where she reports on the technology industry. She joined Reuters in 2023 after working at the San Francisco Standard as a data editor. Tong previously worked at technology startups as a product manager and at Google where she worked in user insights and helped run a call center. Tong graduated from Harvard University.

AI Voice Startup From Ex-OpenAI Researcher Valued at $200 Million

A lead researcher behind OpenAI's voice assistant has raised $40 million in funding from venture firm Andreessen Horowitz for a new artificial intelligence startup focused on helping people have better spoken conversations with computers. WaveForms AI, which launched Monday, is working on AI audio software to pick up on emotional cues and enable more seamless verbal interactions with machines, co-founder and chief executive officer Alexis Conneau told Bloomberg News. The seed funding values WaveForms at $200 million, the company said. The startup is one of a growing number of AI companies moving beyond text-based chatbots to offer human-sounding voice features. The technology offers the promise of creating more versatile digital assistants and customer service agents, among other use cases. At OpenAI, Conneau developed the company's advanced voice mode, which is meant to respond in nearly real time to users' spoken requests with a dynamic, human-sounding voice of its own. Conneau left OpenAI in August and began working on San Francisco-based WaveForms in September -- the same month OpenAI rolled out the voice assistant to all of its paid ChatGPT users.

Investing in WaveForms AI | Andreessen Horowitz

Many companies are focused on building the smartest possible models, which is a crucial goal. There's no shortage of use cases -- AI for law, finance, or medicine, for example - that require the most intelligent models with the best reasoning skills. But for other use cases, EQ is equally (if not more) important than IQ. To create the most immersive and human-like experiences, we need AI that truly feels like talking to a person. This includes giving AI a voice. To that end, we couldn't be more excited to announce our investment in WaveForms AI. The company is training an end-to-end audio language model with the goal of solving the speech Turing test -- creating AI that feels indistinguishable from talking to a human. This is different from today's existing voice models, which typically require multiple steps: transcribing a user's speech, generating a text response, and then running it through a text-to-speech model to respond. WaveForms' new model processes audio natively, enabling more seamless and real-time interactions that fully capture the emotional nuance of a user's voice and respond in kind. We couldn't imagine a better team to tackle this challenge. The company is led by CEO Alexis Conneau, one of the world's leading audio and text LLM researchers and co-creator of OpenAI's GPT4-o Advanced Voice Mode. Cofounder Coralie Lemaitre, a distinguished business strategy leader from Google, will drive product strategy and operations to bring the company's vision to life.

OpenAI, Google veterans launch audio AI startup

Why it matters: While a growing number of chatbots can process voice input, many do so using speech-to-text systems that end up missing important nuance, such as intonation and emotion. Driving the news: WaveForm, which has raised $40 million in a seed round led by Andreessen Horowitz, is headed by Alexis Conneau, formerly of OpenAI, along with Coralie Lemaitre, who previously worked in product strategy at Google. Yes, but: WaveForm, which has just five employees at the moment, is still developing its models, Conneau told Axios. What to watch: Conneau said it's too soon to talk about the specific products WaveForm has in mind, but there should be more details from the company next year.

WaveForms Raises $40 Million for Voice AI Offering | PYMNTS.com

In his time with OpenAI, Conneau developed the company's advanced voice mode, designed to offer real-time replies to users' spoken in a human-sounding voice, the report said. He left the company just before OpenAI introduced a voice assistant for paid ChatGPT users. Now, Conneau is working on an AI that can deliver more immersive voice interactions, which involves creating software that can interpret people's words and how they speak without being trained in advance on the nuances of how they interact, according to the report. For instance, an AI voice assistant may need to register the hesitation in a person's voice in a conversation about a math problem and adapt based on that cue. WaveForms is part of a growing number of AI startups focusing on human-sounding voice features, the report said. "This technology will change our relationship to software," Clarity AI CEO and founder Joan Palmiter Bajorek told PYMNTS in an interview published last week. "For most people, speaking is one of the most natural ways of interacting with another person. So, instead of manually clicking on buttons, interfaces that incorporate voice AI could be some of the most naturalistic we use. Instead of typing a prompt into ChatGPT, you could simply speak your request out loud." As the world becomes more digital, consumers expect voice technology to be more widely available, according to the PYMNTS Intelligence report "How Consumers Want to Live in the Voice Economy." Fifty-four percent of consumers prefer voice technology because it's easier and faster than typing or clicking through websites. Almost half of those surveyed said they think voice tech will be as smart and reliable as humans within five years.