ElevenLabs CEO Predicts Commoditization of AI Audio Models
Mati Staniszewski, co-founder and CEO of ElevenLabs, has indicated that AI audio models are on track to become commoditized in the coming years, providing a strategic insight into the future landscape of audio technology. This statement comes as ElevenLabs continues to focus on advancing audio model architectures.
In a recent interview, Staniszewski emphasized that the company’s research efforts are addressing critical challenges in model architecture, with plans to maintain this focus for the next couple of years. “Over the long term, it will commoditize — in the next few years,” he stated, acknowledging that while variations may exist across different voices and languages, these differences will be marginal.
When questioned about the rationale behind developing models that he anticipates will become commonplace, Staniszewski argued that for the time being, innovative models represent the most significant advantage in the industry. He highlighted that effective and high-quality AI voices are still a pivotal challenge that requires attention: “The only way to solve it is… by building the models yourself,” he explained.
Staniszewski also noted that users seeking reliable and scalable solutions are likely to choose varied models tailored for specific applications. He anticipates a shift towards multi-modal approaches within the next couple of years, where audio and video or audio alongside large language models (LLMs) can be generated simultaneously. He cited Google’s Veo 3 as an illustration of the potential achieved through combined technologies.
Moving forward, ElevenLabs aims to forge partnerships and leverage open-source solutions, blending its audio expertise with that of other technologies. Staniszewski concluded by expressing the company’s commitment to both model development and practical applications in order to deliver lasting impact. He likened their vision to that of Apple’s synthesis of hardware and software, suggesting that a combined approach to product and AI will yield the most innovative use cases.
