"Over the long term, it will commoditize - over the next couple of years,"
"Even if there's differences - which I think will be the truth for some voices, some languages - on its own, the differences will be smaller."
"The only way to solve it is... building the models yourself, and then, over the long term, there will be other players that will solve that, too,"
"So, you will create audio and video at the same time, or audio and LLMs at the same time in a conversational setting,"
ElevenLabs has solved several model architecture challenges and will continue focusing on audio model development for the next one to two years. AI audio models are expected to become commoditized over the next couple of years, though some differences will remain for particular voices and languages. Building proprietary models provides the largest immediate advantage for improving voice quality and interaction naturalness. Current audio quality problems require in-house model work to solve. Reliable, scalable deployments will likely use different models for different use cases. An increasing number of models will adopt multimodal or fused approaches combining audio with video or language models. ElevenLabs plans partnerships and engagement with open-source technologies.
Read at TechCrunch
Unable to calculate read time
Collection
[
|
...
]