Deepgram Acquires Daily to Enhance Voice AI Infrastructure, Keeps Pipecat Open Source

Deepgram announced on Thursday its acquisition of Daily, the company behind Pipecat, a popular open-source voice AI infrastructure. This move aims to bolster Deepgram's capabilities in the voice AI sector, with a commitment to keep Pipecat open-source.

Pipecat is a framework designed to manage voice and video streams, integrating elements like speech-to-text (STT), large language models, and text-to-speech (TTS) models while minimizing latency. Daily, its maker, provides hosted infrastructure for various voice and video calls. The acquisition is expected to provide Deepgram with a deeper understanding of real-time voice and video infrastructure.

Daily has secured over $62 million in funding from investors including Tiger Global, Highfield Capital, Nat Friedman, Lachy Groom, and Kevin Weil. Deepgram itself achieved unicorn status earlier this year following a $130 million Series C funding round. Notably, both companies were part of the same Y Combinator batch in 2016.

Kwindla Kramer, Daily’s co-founder, stated that the Pipecat framework emerged from experiments with AI-to-human communication over voice and text around the time GPT-4 was released. Kramer emphasized his preference for Deepgram due to CEO Scott Stephenson's commitment to maintaining Pipecat as open-source "forever" and the alignment of engineering cultures. He noted, "Scott was super committed to Pipecat being open-source and vendor-neutral forever. I think there is both a personal commitment and commercial driver there."

Deepgram CEO Scott Stephenson explained that the company had long sought to engage with open source, identifying an orchestration framework for agents as a key area. "We want to open up the ecosystem so everybody can build voice agents," Stephenson remarked, encouraging developers to use "the best models for your use case," including partners, while Deepgram works to enhance its own offerings.

Stephenson added that the acquisition will enable developers to build expressive agents and end-to-end speech systems by integrating different models and a context bus. The deal also allows Deepgram to explore applications beyond voice through Pipecat, which caters to real-time communication across various formats. Kramer highlighted, "We never thought about Pipecat as a voice toolkit. We thought about it as the toolkit we needed to build this new kind of AI-native software."

In conjunction with the acquisition, Deepgram is launching the Real-Time Context Standard (RCS), an open protocol co-developed with the Pipecat team. RCS defines metadata such as audio context, instructions for language models on signal usage, and guidance on response acoustics.

The voice AI sector is experiencing significant growth and consolidation, with both model developers like ElevenLabs, Gradium, and Fish Audio, and infrastructure providers such as LiveKit and Vapi, securing substantial investments.

What to watch: The evolution of open-source voice AI infrastructure and new agent development.

Editor's note: The draft is comprehensive, well-structured, and incorporates all key facts and quotes from the provided text.