Mars: Speech recognition shifted from Gaussian mixture models to deep neural networks
Jason Mars · Jason Mars · Jul 28, 2017 · at 6:39
Jason Mars, computer science professor at University of Michigan and co-founder of Clinc, outlines the industry-wide architectural transition in speech recognition infrastructure at an a16z roundtable.
0:00 / 0:08exact quote · 8.3s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“Google used to use Gaussian mixture models in the early days, and now everyone has moved to deep neural networks. To do the speech recognition, ah, task.”
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 Jason Mars
Mars: AI assistant infrastructures are monopolized by major tech companies
“These kinds of infrastructures are monopolized essentially by companies like Apple, Google, Microsoft, and so forth.”
Mars: Converting search to AI assistants requires 100x data center scaling
“If you moved all of the queries that were search queries and turn them into intelligent personal assistant queries, we would have to grow our data centers by two orders of magnitude.”
Mars: Apple requested Sirius AI project rename, resulting in Lucida
“After Sirius gained a good bit of traction, Apple reached out to us and said, you probably should change the name. And so we're now calling it Lucida.”
Mars: Porting Sirius to GPUs and FPGAs yields 10x speedup
“We can get significant speedups when we port these algorithms To GPU, when we leverage GPUs and FPGAs to build future servers, and what that means is we can bring that scalability gap down, and so we can get about 10 X back if we move by taking these, ah, by t…”
Mars: Tech giants are building future data centers around deep learning
“Deep learning is absolutely the bet that companies are making, right? So Google has put a lot of resources into deep learning you know, Facebook. We're designing future data centers and future infrastructures around this one technique, and it's primarily becau…”