Meta Platforms has acquired WaveForms AI, a speech recognition startup founded by former OpenAI and Google researchers, in a deal that could strengthen the social media giant’s artificial intelligence chatbot capabilities.
The Information reported the acquisition Thursday, though financial terms remain undisclosed. WaveForms raised $40 million in December at a $200 million valuation, suggesting the purchase price likely reached hundreds of millions of dollars.
Veteran AI Talent Joins Meta
WaveForms founders Alexis Conneau and Coralie Lemaitre have joined Meta as part of the deal. Conneau previously worked at OpenAI on GPT-4’s Voice Mode feature, which enables voice-based interactions with the language model. Lemaitre brought advertising strategy experience from Google.
The startup, launched in 2024, was developing what Andreessen Horowitz described as an “end-to-end audio language model.” This approach consolidates three traditionally separate AI functions—speech transcription, response generation, and text-to-speech conversion—into a single model.
Streamlined Speech Processing Technology
WaveForms’ technology represents a significant efficiency gain over conventional speech AI systems. Rather than requiring separate models for each processing stage, the company’s approach allows one model to handle the entire speech-to-speech pipeline.
The startup also aimed to incorporate emotional cue detection, enabling its model to adapt responses based on the speaker’s emotional state. At the time of its funding round, Bloomberg reported that WaveForms had not yet released a product.
Second Voice AI Acquisition in Weeks
This marks Meta’s second voice AI acquisition within a month. The company previously purchased Play AI, which developed voice agent platforms for customer support applications, after the startup raised $21 million.
According to internal memos, Meta plans to integrate Play AI’s technology to enhance its Meta AI chatbot’s voice capabilities. The chatbot currently supports voice commands similar to ChatGPT and is being developed to translate speech in Instagram Reels videos.
Research Unit Expands AI Capabilities
The acquisitions support Meta’s broader AI research efforts through TBD Labs, which houses many researchers recently recruited from OpenAI and Google. This unit is reportedly developing new Llama models, including versions designated internally as Llama 4.5 and Llama 4.X.
These upcoming models focus on improved reasoning capabilities, though Meta has not announced release timelines or confirmed whether they will maintain the open-source licensing of previous Llama versions. Many current Llama models already process voice input, making WaveForms’ expertise particularly relevant for future developments.
The WaveForms acquisition continues Meta’s strategy of acquiring specialized AI startups and their talent to compete with rivals like OpenAI and Google in the rapidly evolving artificial intelligence landscape.