Meta Acquires WaveForms AI Voice Startup to Boost Superintelligence
Meta Platforms has continued its aggressive expansion into the artificial intelligence frontier with the acquisition of WaveForms, an AI audio startup specializing in hyper-realistic and emotionally intelligent voice technology. The undisclosed deal marks Meta’s second significant AI audio acquisition within a month, underscoring a strategic push to bolster its newly formed Superintelligence Labs division.
WaveForms, founded just eight months ago in December 2024, quickly made a name for itself in the burgeoning AI voice sector. The San Francisco-based startup successfully raised $40 million in seed funding earlier this year in a round led by prominent venture capital firm Andreessen Horowitz, which valued the company at $200 million. At its core, WaveForms aims to solve the “speech Turing test,” a formidable challenge in AI where a listener cannot distinguish between human speech and AI-generated voice. Their technology focuses on developing “Emotional General Intelligence” (EGI) through audio large language models capable of interpreting and replicating subtle emotional nuances in human speech. This sophisticated capability allows for more seamless, real-time, and emotionally resonant interactions with AI.
The startup was co-founded by Alexis Conneau, a distinguished researcher who previously contributed to audio research at Meta and played a pivotal role in OpenAI’s GPT-4o Advanced Voice Mode, and Coralie Lemaitre, a former advertising strategist at Google. Both Conneau and Lemaitre are set to join Meta’s Superintelligence Labs, integrating their expertise directly into the tech giant’s ambitious AI initiatives.
This acquisition follows closely on the heels of Meta’s purchase of PlayAI in July, another voice AI startup focused on generating human-like voices, including multilingual voice cloning and real-time speech synthesis. The entire PlayAI team, comprising around 35 individuals, joined Meta and now reports to Johan Schalkwyk, a former Google speech AI researcher who recently became the Voice Lead within Superintelligence Labs. These consecutive acquisitions highlight Meta’s concerted effort to close any perceived gaps in its AI audio capabilities, especially as it reportedly works towards integrating real-time, two-way voice conversations into its AI chatbot.
The strategic importance of WaveForms and PlayAI is deeply intertwined with Meta’s overarching “personal superintelligence” vision, articulated by CEO Mark Zuckerberg. This vision aims to develop AI systems that not only surpass human cognitive abilities but are also deeply personalized and accessible to everyone, potentially via devices like Meta’s AI smart glasses. To achieve this, Meta has embarked on an unprecedented talent acquisition spree and a massive infrastructure build-out. The Superintelligence Labs, officially launched in July, consolidates Meta’s core AI teams under the leadership of Alexandr Wang, the former CEO of Scale AI, whom Meta brought on as Chief AI Officer after a $14.3 billion investment in his company.
Within Superintelligence Labs, a specialized unit known as “TBD Lab” is already developing Llama 4.5, the next generation of Meta’s language model, with a focus on enhancing reasoning and agent skills. Meta is pouring billions into this endeavor, recruiting top AI researchers from rivals like OpenAI, Google, and Apple with lucrative compensation packages, and constructing colossal AI data centers, including the “Prometheus” 1-gigawatt-plus supercluster. This aggressive strategy signals Meta’s pivot away from directly competing with productivity-focused AI tools, instead prioritizing AI for entertainment, social connections, and lifestyle features, while also aiming to automate its entire advertising workflow by 2026.
The acquisition of WaveForms, with its pioneering work in emotional voice AI, positions Meta to create more empathetic and natural AI interactions across its diverse platforms, from social media to the metaverse, ultimately aiming to deepen user engagement and redefine the human-AI communication landscape.