Chan: ML audio scoring lets music teachers instruct multiple students simultaneously
Connie Chan · What's Next for Education Startups in 2019 (Part II) · Feb 2, 2019 · at 10:07
Connie Chan, General Partner at Andreessen Horowitz, explains how Chinese EdTech platforms like VIP Peilian leverage machine learning for music instruction.
0:00 / 0:38exact quote · 38.6s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“And then not only is that the case, they use the machine learning aspect to help the teachers with scoring the kids and scoring the performance, because with music, just like with language, there is a actual pitch, there is an actual tempo, an actual rhythm that you're supposed to play. Yes. Right, so they can take the composition score and then hear your actual performance and give you a grade, which then allows One teacher to teach two or three students at the same time, which then unlocks even more cost savings, allowing more parents to give their kids these music lessons that they would typically not be able to afford.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Connie Chan
Chan: Massive market sizes justify venture funding for competing startups
“Because the market's so big The pie is so large that it does actually warrant that crazy risk taking, that even if a company is already doing well, I might just give a ton of money to another competitor because if they win, the pie is so much bigger.”
Chan: EdTech startups should build mobile apps before websites
“This is probably a contrarian view, even in Silicon Valley, but I would build for mobile first, and I would build your app before you build your website, because it will drastically unlock different ways of thinking.”
Chan: Upfront subscriptions are a flawed business model for EdTech
“For this category, I don't think subscription is necessarily the best model. Because for me to sign up for a subscription, I have to think I'm going to take more than one class, right? I mean, why not instead let me pay per course, and for other courses, if yo…”
Chan: Chatbot adoption failed due to missing use cases, not discovery
“My concern is that the issue around bot usage and adoption is not actually the discovery piece. It's that the great use cases for the bots haven't been created yet. And that's more a function of Facebook educating developers or developers coming up with the ri…”
Chan: Photo tools benefit from unbundling, social networks require single apps
“It makes sense for a photo editing company to break that up into multiple apps because from a user metric standpoint, I will download four or five of them. But for a company like Facebook, I only want one if it's the same exact network of people.”