@build
Conversations with founders, engineers, investors building great products and companies. Hosted by @arish
The Nuance Acquisition - what it means for AI and Human Speech technologies
Microsoft's acquisition of Nuance earlier this week was notable for two things. One was definitely the price tag almost $20 billion for a company at a one $4 billion run rate. That's a very large multiple. The second was a light it shared on the importance of healthcare and AI in the future. As per Senatela, the CEO of Microsoft, Ali is technology's most important priority, and healthcare is its most urgent application
To discuss this topic further, I have invited William Carmichael, who is on the product team at Deepcram, a company that develops Ali model for speech. William, welcome to Swell. I would love to ask you two questions. The first one is, what did you think? The the the the Nuance acquisition, what it for the industry. And the second one is what is the current state of this Peach and Ali industry? What exciting innovations do you see happening in the future? There's
Will Carmichael
@Will.C · 1:54
Thanks for the invitation into both Swell and this conversation, Arish. I think this acquisition, obviously the price tag is what sticks out the most to me. First of all, it's very, very good kind of reinforcement of the value of voice and where it's headed in the industry. My CEO, Scott Stevenson, actually said it best when he said just like how GM when they acquired Cruz, boosted the industry and flooded it with money
Thank you Will, for your reply and the context you provided. I think your CEO was spot on with the GM Cruise example. I think Microsoft can definitely do something similar in terms of providing any scale and investments that are needed in such core technologies to really grow the whole industry and take it forward. So I'm looking forward to those developments for sure. And as the space grows, as the industry grows, I think a lot of innovation will be happening for you
Mark Smithivas
@msmithivas · 0:39
Hi. I just wanted to make a quick comment that I think this whole field is fascinating and opens up a lot of use cases. As you say, I'm very curious what a lot of these platforms that that we're on now, like Swell Clubhouse, Twitter spaces. They're all audio based. And I wonder if behind the scenes they are training some kind of AI model to recognize what people are saying. So I'd be curious to know your thoughts on that
Sudha Varadarajan
@sudha · 1:11
And so translation is a key area in which we would love to explore the use of AI, but that's something very far out in the future. Accessibility. We want Swell to be as inclusive as possible. So we do believe that AI can help play a part in ensuring that we can make Swell more accessible to playing people who have disabilities. So these are some use cases we have in mind, but these are very far out in the future, and we will continue monitoring developments in this space
Bernie Goldbach
@topgold · 2:42
And knowing that one of them from Amazon understands the voice register of my son nine years old, talks to him by his name, lets him put specific reminders for him on the smart devices is so amazing as a nine year old, knowing he's going to go through life, expecting his car to talk back to him, expecting the lights to come on by command, by voice command
kapulono
@kapulono1 · 3:36
So yeah, you know, concerned and not quite sure yet of where the technology will take us. But hey, who's to say we will not need translators, interpreters language teachers anymore in the future? Or how will they coexist? Maybe these tools will be used to the advantage of educators
Ryan Cartier
@Cartier · 2:00
On your comments about different platforms and how they're using voice. Keep in mind that even though a platform might not be used voice, anything that's online and accessible can use voice. So, for instance, like you already know that Google and Alexa are recording your voice, and then people listen to the conversation and they use that to improve their product. And that's something that not everyone's excited about
Ryan Cartier
@Cartier · 3:04
I keep hitting the person's particular message to reply to it, but it doesn't add their tag to indicate to them. I've replied to them specifically, so I would love that feature. This is to add capital loan when you're talking about language. So in addition to people learning languages for work purposes, a lot of people just learn languages because it's fascinating or they happen to be in another country driven by sheer use
Hi, Bernie. Yes. The developments are so fascinating in this space. I have always believed that the best interface is no interface when the technology is invisible and works for you without having to click and type and scroll and and tap and that turn a tiny screen with squinted eyes. So I think voice based interfaces can take us closer to that kind of a screen free world where the technology is there. It's powerful, but it's invisible, and we don't have to fight it's
If you long press on any one of these peach bubbles, it gives you a reply option, which automatically fills the handle of the person you're replying to, so I hope that helps coming back to the conversation itself. Personally, I love the idea of learning different languages. I feel the more languages people can learn, the better it is. The wire your brains differently and really open up your perspective and outlook. However, purely from a tech perspective. I agree with Ryan
Arun Manche is the founder of ihero, which is a startup focused on building voice data models for people with speaking disabilities. I have been doing an interview with him on the Bell's Forecast and I will add a link below if you want to listen to that. I'm also inviting Arun to join this conversation and share his thoughts on the Nuanced acquisition
Deborah Pardes
@DBPardes · 2:35
Then we see itself correct. And then we sort of say, this is okay. This is part of our world. But I'm fascinated by the fact that we're talking on an app that, for me, helps me feel more secure, more trusting. And yet when we fold this into innovation and progress, there will be points where the cleverness of AI to understand the nuance of the human voice will also be our pause. We'll pause for a second and say, Wait a second
Ryan Cartier
@Cartier · 4:07
And when you walk into someone's home, people buy these Alexa, the Google thing, and they take advantage of it for its utility and interact with the voice, after which they are surprised to find out that it's recording their voice. People at Amazon listen to it, and then they try to figure out how to improve the product. So the fact that the adoption is so far ahead of the question whether or not this is a good idea
Deborah Pardes
@DBPardes · 1:33
And I'm wondering if anybody in the thread knows about the progress there, and if there's any papers that have been written about reasons to keep it limited, or have there been leaps in scientific method around it, and if they can actually map the human sequence so much that they can actually clone it completely? I'm hoping not. But you just remind of that whole episode in our history
Arun Munje
@Arun · 1:40
I've been listening to some excellent points made about covering this topic in from many different angles. I think that usually with every new innovation, there is always is a moral ethical question that needs to be addressed, and usually the answer to those are very complex and easily become a news and conversations. For now. At this point I'll focus on the positive part of it for now about the speech to text in general
Ryan Cartier
@Cartier · 4:20
Also, perhaps a little bit more related to your question. What's going on now that we are not aware of? I have a video. I'll add the link, but this is where I'm curious. Does it derail things too much if I go in a totally different direction? So answer me on that as well. So this is a video I saw ten years ago about a liver, no pancreas kidney
Ryan Cartier
@Cartier · 3:01
Let me correct myself. It is not the cells that shorten. It is the telomeres, and they're the caps at the ends of chromosomes that produce protect DNA from damage, which shorten over time as we get older and eventually it gets to an unhealthy state. And they also note that these short telomeres can be used to predict the early onset of diseases. Now there is an enzyme in our cells called telomerase, if that is the correct pronunciation that protects the telomeres and rebuilds and lengthens them
