Showing posts with label Speech recognition. Show all posts
Showing posts with label Speech recognition. Show all posts

Thursday, November 01, 2012

Google Also Leads Voice Search: No Surprise

Image representing Google as depicted in Crunc...
Image via CrunchBase
When Siri came out there was talk finally Google has competition. I never bought into that. The front end might have been cute, but Apple and search? Come on.

Why I Won’t Be Using Google’s New iPhone Voice Search

For me it is less about moving from typing text to voice and more about moving from language to language to language.

That would be the faster way to wipe out illiteracy from the planet. If voice inputs and outputs can prove to be almost sufficient a lot of knowledge could move around right away.

Google offers up secret sauce on new voice search
this app gives Siri a run for her money. It is lightening fast, has a clean layout, and gives highly accurate results
Google explains how more data means better speech recognition
More data helps train smarter models, which can then better predict what someone say next ...... more data trumps better algorithms ..... For the voice search tests, the Google researchers used 230 billion words that came from “a random sample of anonymized queries from google.com that did not trigger spelling correction.” However, because people speak and write prose differently than they type searches, the YouTube models were fed data from transcriptions of news broadcasts and large web crawls. ...... As consumers demand ever smarter applications and more frictionless user experiences, every last piece of data and every decision about how to analyze it matters.
Large Scale Language Modeling in Automatic Speech Recognition
The n-gram approach to language modeling (predicting the next word based on the previous n-1 words) is particularly well-suited to such large amounts of data: it scales gracefully, and the non-parametric nature of the model allows it to grow with more data. For example, on Voice Search we were able to train and evaluate 5-gram language models consisting of 12 billion n-grams, built using large vocabularies (1 million words), and trained on as many as 230 billion words.
Enhanced by Zemanta

Wednesday, November 09, 2005

Memo To Bill Gates



A memo from Gates has been leaked where he says Microsoft is "at risk" from Google. I figured I would respond, so here is me composing a memo to that other Bill from the 1990s.

Mr. Software Architect.

Part of your problem is simply ageing. There was IBM, and then Microsoft came along, and Microsoft eclipsed IBM itself in market capitalization. You might be IBM, and Google might be Microsoft. Empires come and go. So at some level, just make peace.

I am a huge fan of your foundation though. I wish you were 10 times richer, I am so impressed with your work for health care in the poor countries. And of course you are a terribly smart, creative guy. I am easily a fan.



At some point I think a company like Microsoft should just plough in all that extra cash into becoming a venture capitalist firm, or at least growing a wing in that direction, I think. It is young scientists and the entrepreneurs who come up with the cuttinge edge ideas, or at least in most cases.

I think your problem is that you thought you woke up to the internet in 1995, and you did not. Then you thought you did it in 2000, and you did not. Now you think you are doing it in 2005, and you are not. For good or ill, Microsoft remains a Windows company. Microsft never really became a dot com.

But if Microsoft were to reinvent itself, what might it do? Here are some suggestions I offer.
  • You don't have to ditch Windows outright, but shift focus to the online world. That is the present and the future. Down the road, Windows either disappears, or becomes invisible.
  • Could you take word processing online, and could you make it ad-based? Do you even want to?
  • Could you take the lead on becoming a digital publisher? License Google's ad program if you have to, if you can't replicate it. But noone is taking publishing online. Maybe you can take the lead. All books - textbooks, fiction, non-fiction - should go online and be free, as in ad-based. Could you take the lead on that one? You are the leader in word processing offline. Could you go online?
  • You have had some interesting thoughts on speech recognition technology in the past. What is the progress there? Keep working there. Noone seems to be competing with you there. Down the line people should be able to talk to their computer in any language.
  • Put as much work into your browser as you have been putting into your Windows and Office. Because W and O are passe. The browser will go far.
  • You have done good with non-PC devices. Maybe you should work on the software for "free" cellphones that will work in a citywide soup of wireless broadband.
  • The internet and wireless broadband are not one and the same thing, just like webpages and blogs are not one and the same. Don't get too hung up on the internet. Think wireless broadband.
  • Expand your facilities in India. It's not just about cheap, smart engineers. India is going to be a huge market on its own.
  • Google itself has said there is room for more than one player in the search market. Search is the center piece of the Google magic. Get in the action. Keep working at it.
  • See if you can imitate Google. They like to offer web services that are online and ad-based. If those two parameters are too narrow for you, you are not competing with Google. Wean yourself away from the habit of asking people for money directly.
More later.

December 7: Microsoft To Invest $1.7 Billion In India (BusinessWeek)

Reblog this post [with Zemanta]