Suggestions
Dylan Fox
Founder/CEO - AssemblyAI
Dylan Fox is a highly accomplished professional with a diverse background in business administration, economics, public policy, and software engineering.
He holds a degree in Business Administration with a focus on Business Economics & Public Policy from The George Washington University. Prior to this, he studied at Eastern Regional High School.
Throughout his career, Dylan has held key roles in notable organizations. He served as the Founder & CEO at AssemblyAI, showcasing his leadership and entrepreneurial skills. Additionally, he contributed his expertise as a Sr. Software Engineer specializing in Machine Learning at the Emerging Products Lab in Cisco.
His entrepreneurial spirit led him to establish YouGive1, where he served as Founder and CEO, demonstrating his ability to drive innovation and create successful ventures. Earlier in his career, Dylan also gained experience as a Lead Research Assistant at NDP Consulting.
Dylan's educational background and extensive experience in both business and technology have equipped him with a unique skill set that blends strategic thinking, technical proficiency, and business acumen. His contributions to various organizations underscore his ability to lead, innovate, and drive impactful projects across different sectors.
Highlights
Who's at Google Next this week?? Ping me if you want to try to connect in person out here! Especially if you're building with the AssemblyAI API 😁
Very excited to release Universal-1 today -- our newest model for AI speech-to-text.
Some key stats:
- Universal-1 is trained on 12.5 million hours of multilingual audio data
- 72% preferred Universal-1 to our most recent model, Conformer-2, in human evals
- 71% better speaker count estimation and 14% better word timestamp estimation compared to our prior models
-
22% more accurate than APIs from Azure/AWS/Google and 13.5% more accurate than models like Whisper, with up to 30% fewer hallucinations
- Just 38 seconds to process 1 hour of audio
There is more voice data being created on the internet today than ever before. Our goal is to build the best API platform for developers and organizations to transform and understand all of this voice data with the help of AI.
Speech AI systems have gotten a lot better over the past 2 years, but there is still a lot more work to be done in order to create Speech AI systems that are more context aware, lower cost, faster, and more capable. Today's release is another major milestone in our journey towards achieving this goal, and I am super proud of what our team was able to accomplish here.
To quote Domenic Donato, more than a model... Universal-1 is a technological achievement unlocking large-scale data processing and high-performance computing infrastructure at AssemblyAI, enabling us to create better and better Speech AI models for our users going forward.
It is still Day 1 here at Assembly, and we have a lot more planned for the coming months!
In the meantime, you can read all about Universal-1 in our technical overview below 👇
