Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Saturday April 22 2017, @01:28AM   Printer-friendly
from the His-Master's-Voice dept.

Submitted via IRC for AndyTheAbsurd

Amazon.com Inc's (AMZN.O) chief technology officer is working toward a day when people can control almost any piece of software with their voice.

The company on Wednesday rolled out the technology powering Alexa, its voice assistant that competes with Apple Inc's (AAPL.O) Siri, to developers so they can build chat features into their own apps, CTO Werner Vogels said in an interview. The service, Amazon Lex, was in a preview phase since late 2016.

[...] Processing vast quantities of data is key to artificial intelligence, which lets voice assistants decode speech. Amazon will take the text and recordings people send to apps to train Lex - as well as Alexa - to understand more queries.

That could help Amazon catch up in data collection. As popular as Amazon's Alexa-powered devices are, such as Echo speakers, the company has sold an estimated 10 million or more. Apple has sold hundreds of millions of iPhones and other devices with Siri.

[...] As with other cloud-based services, Amazon will charge developers based on how many text or voice requests Lex processes.

Source: https://finance.yahoo.com/news/amazon-rolls-chatbot-tools-race-181632556.html


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 1) by anubi on Saturday April 22 2017, @06:32AM (2 children)

    by anubi (2828) on Saturday April 22 2017, @06:32AM (#497825) Journal

    I was hoping for more refinement and specialized speech processor chips that were very speaker dependent... so that if I spoke, the computer would recognize both the phrase and it was me speaking... if someone else spoke the exact same thing, unless they sounded damned similar to me, it would be unrecognized.

    I have seen some processors for Arduino that seem to be speech recognizers, but I have yet to mess with any. Anyone else mess with any of 'em?

    Intended use is to monitor my house... especially my bedroom ... and put my system into a defensive mode when so ordered by as much as saying a particular phrase. A few high voltage discharges ( neon-sign transformer / Jacob's ladder kinda stuff ) can be very visually intimidating and I think could likely cause an intruder to reconsider hanging around to see if he can find the human-sized bug zapper. ( Yeh, I know... best run it from a UPS. ).

    Psy-ops. Hollywood style.

    The other intended use is to throw the powertrain control module I am building for my van into various imminent shutdown modes should I utter a phrase like "please don't hurt me!". The system will then follow instructions - that unless countered by later phrases - will do things like hijack the transmission controls and shut down the fuel pump, causing the engine to pull vapor into its high pressure hydraulic fuel injectors, which will then require about a thirty minute to one hour bleeding procedure to be done before the engine can be restarted. Even they get the engine going again, with the PCM in control of the transmission solenoids, I can have the PCM purposely lock up the transmission so just as they try to put it into gear, it will stall out the engine. It will take them so long to figure out what I did that they would probably much prefer finding another target to mess with.

    --
    "Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
  • (Score: 2) by PiMuNu on Saturday April 22 2017, @09:27AM (1 child)

    by PiMuNu (3823) on Saturday April 22 2017, @09:27AM (#497850)

    What would a speech processor do differently to a conventional CPU?

    • (Score: 1) by anubi on Sunday April 23 2017, @05:49AM

      by anubi (2828) on Sunday April 23 2017, @05:49AM (#498182) Journal

      This whole thing is supervised by an Arduino. Selected for its trustworthiness, its simplicity, as well as its miserly power consumption... especially when sleeping.

      But an Arduino loses terribly on raw processing power.

      So, I offload as much stuff as I can onto dedicated sub-processors.

      Intention is failure of subprocessing systems will not lock up the Arduino. Just trip failure flags.

      Nor can one malfunctioning subsystem affect other subsystems.

      Another thing is I need very high levels of task fidelity. Meaning each subprocessing unit has a particular thing its watching. In my case, the most time-critical thing is two variable reluctors telling me how many gear teeth have passed on the engine drive crankshaft, as well as the differential rear end. I want to count every last one of 'em. They are used to sense speed, engine tach, transmission slippage, fuel consumption, mileage, engine load, and possibly more.

      --
      "Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]