Stories
Slash Boxes
Comments

SoylentNews is people

posted by LaminatorX on Friday March 13 2015, @11:54PM   Printer-friendly
from the Hello-Computer dept.

An open-source computing system you command with your voice like Apple's Siri is designed to spark a new generation of "intelligent personal assistants" for wearables and other devices. It could also lead to much-needed advancements in the datacenter infrastructure to support them.

Sirius, built by University of Michigan engineering researchers, is similar to Siri, Microsoft Cortana and Google Now—robust applications that accept voice instructions and questions, interpret them and answer in spoken words.

Sirius even uses many of the same "fancy algorithms," said Jason Mars, U-M assistant professor of computer science and engineering and co-director of Clarity Lab where Sirius was developed. But unlike its expensive and locked-down commercial counterparts, Sirius is free and can be customized.

http://phys.org/news/2015-03-siri-open-source-digital.html

[Paper]: http://jasonmars.org/wp-content/papercite-data/pdf/hauswald15asplos.pdf

[Source]: http://ns.umich.edu/new/multimedia/videos/22734-build-your-own-siri-an-open-source-digital-assistant

This discussion has been archived.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: 2) by physicsmajor on Saturday March 14 2015, @12:23AM

    by physicsmajor (1471) on Saturday March 14 2015, @12:23AM (#157589)

    The word "open" appears in the article PDF no less than 27 times. The press release outright states you can build your own version. I'm excited, I'm down to try this out. So... how can I do that? The paper includes no source, no link to source, or any mention of how this will be released or under what kind of license. A link to a working demo on AWS instead of a screenshot from a workstation would have gone a long way, if they're still cleaning up before public source release.

    It looks like they've done a lot of work. I'm ready to commend them for it once they actually stand behind their claims. But they need to actually open this up, not just give a lot of lip service to the idea.

    • (Score: 5, Informative) by Anonymous Coward on Saturday March 14 2015, @12:37AM

      by Anonymous Coward on Saturday March 14 2015, @12:37AM (#157595)

      > Now where's my GitHub link?

      https://github.com/jhauswald/sirius [github.com]

      Took me less than 60 seconds to find it, with a single google search [google.com]and click on the first link. I spent more time writing this post.

      • (Score: 2) by hendrikboom on Saturday March 14 2015, @03:13AM

        by hendrikboom (1125) Subscriber Badge on Saturday March 14 2015, @03:13AM (#157642) Homepage Journal

        So does this mean we now can have free, open-source voice recognition? Or does this somehow bypass the voice-to-text stage?

        • (Score: 1, Informative) by Anonymous Coward on Saturday March 14 2015, @03:22AM

          by Anonymous Coward on Saturday March 14 2015, @03:22AM (#157646)

          > So does this mean we now can have free, open-source voice recognition?

          Already available: CMU Sphinx [wikipedia.org]

          > Or does this somehow bypass the voice-to-text stage?

          Sirius isn't about the voice recognition, its about the natural language processing in order to figure out the intention behind the words.

        • (Score: 0) by Anonymous Coward on Saturday March 14 2015, @03:41AM

          by Anonymous Coward on Saturday March 14 2015, @03:41AM (#157653)

          From the article, "Speech recognition came from Carnegie Mellon University's Sphinx,"

        • (Score: 2) by TheRaven on Tuesday March 17 2015, @08:48AM

          by TheRaven (270) on Tuesday March 17 2015, @08:48AM (#158789) Journal
          Speech recognition comes from two open source projects, one from CMU and one from Microsoft Research. All of the components that they've assembled are existing open source projects, their contribution is assembling them into something that works sort-of like Siri. It's mostly about providing a test workload for people doing research on datacentre operating systems and architectures.
          --
          sudo mod me up
    • (Score: 3, Funny) by nightsky30 on Saturday March 14 2015, @05:45PM

      by nightsky30 (1818) on Saturday March 14 2015, @05:45PM (#157813)

      It can't beat this!

      https://isotropic.org/papers/chicken.pdf [isotropic.org]

  • (Score: -1, Offtopic) by Anonymous Coward on Saturday March 14 2015, @01:26AM

    by Anonymous Coward on Saturday March 14 2015, @01:26AM (#157616)

    Will be deeepreeciated, coopted, rewritten from scratch by systemd type faggots, so why bother?

  • (Score: 2) by TLA on Saturday March 14 2015, @02:53AM

    by TLA (5128) on Saturday March 14 2015, @02:53AM (#157633) Journal

    Siri took out an injunction on me for stalking.

    --
    Excuse me, I think I need to reboot my horse. - NCommander
  • (Score: 3, Interesting) by buswolley on Saturday March 14 2015, @06:00AM

    by buswolley (848) on Saturday March 14 2015, @06:00AM (#157672)

    This story might not get the comments it deserved. I'd love if it were reposted on Monday morning

    --
    subicular junctures
    • (Score: 2) by VLM on Saturday March 14 2015, @01:50PM

      by VLM (445) on Saturday March 14 2015, @01:50PM (#157754)

      I would theorize that its something that is not in much demand or there would have been CLI interfaces in emacs doing the same thing for the last 30 years. I suppose there's always meta-x eliza mode which was pretty entertaining on 80s home computers.

      Something that sells phones is not necessarily interesting. As an example of the general class of problem, look at Apple releasing a "the new connector to bind them all forever" roughly annually. Theres too many to count now and none of them are very interesting. Oh this one uses magnets, this one is an overgrown USB, yet another display connector, this one has 30 pins, this one can be flipped upside down and still work, whatever. Another example is the truly weird apple fixation on thin phones and laptops, as if ANYONE cares how thin their laptop or phone is (weight, sure, but who cares how thin it is?). Its just not interesting to anyone unless the presenter is wearing a black turtleneck on stage.

      • (Score: 3, Interesting) by TheRaven on Saturday March 14 2015, @04:50PM

        by TheRaven (270) on Saturday March 14 2015, @04:50PM (#157800) Journal
        The work is being presented at ASPLOS on Monday. Those of us who are there will be better able to comment intelligently afterwards...
        --
        sudo mod me up
  • (Score: 2) by Gaaark on Saturday March 14 2015, @04:41PM

    by Gaaark (41) on Saturday March 14 2015, @04:41PM (#157798) Journal

    Now i need someone with skills to build me a .deb file, or even a easy to use .tar.gz...

    I used to have some skills, now my time is spent and my memory is sh*t: installed plex on my laptop (gonna try to stream to a roku 2 so my wife and i can watch the shows on my external harddrives on my old tv):

    :trying to get the external harddrives recognized by plex was a bitch and i actually had to think again... it was kind of nice, really! :)

    Long time spent getting all those files into plex, though, and it identified the bbc show 'Heartbeat' as Police Academy 7 for some reason, lol.

    Will see if my laptop is powerful enough to run the shows/movies at a reasonable clip: if so, will buy the roku!

    Something like sirius would be nice, but not sure i could do a git build: getting old sucks.

    --
    --- Please remind me if I haven't been civil to you: I'm channeling MDC. ---Gaaark 2.0 ---
    • (Score: 3, Interesting) by NotSanguine on Saturday March 14 2015, @08:13PM

      An interesting point.

      Not sure if this is a stable release, but it sure doesn't build cleanly. Which isn't really very surprising since according to the University of Michigan announcement [umich.edu]:

      To make Sirius, the researchers stitched together four established open-source projects that they say rely on techniques and algorithms that resemble those in commercial systems.

      Speech recognition came from Carnegie Mellon University's Sphinx, Microsoft Research's Kaldi and Germany's RWTH Aachen "RASR" project. The question-and-answer system came from OpenEphyra, which laid the foundation for IBM's Jeopardy-winning Watson. Image recognition came from the SURF computer-vision algorithm behind Swiss tech entrepreneur Herbert Bay's company Kooaba, which was recently acquired by Qualcomm.

      Based on my (very limited) experience so far, "stitched together" may be too strong a term.

      Dependency checks in various components are all over the place and non-existent in some cases. Some of the source sub-trees are set up nicely with autoconf and others not so much. This is very much a work in progress.

      If you wish to build this project (I'm trying to do so), it's going to take a while to get everything straight.

      N.B. I am not a dev guy, but I am an infrastructure guy with long linux/unix experience and a good understanding of dev tools and building from source and minimally hacking to get stuff to compile and (hopefully) run.

      I'll get it done pretty soon (I believe I've addressed most of the dependency issues, although I did see some hard errors complaining about specific gcc versions. grrr!), assuming I don't run into too many other issues.

      This project is nowhere near ready for a guy like me (I'll get there eventually, but it's been a pain so far and will continue to be so, methinks), so it's definitely not ready for any sort of real deployment or packaging into a production distribution.

      I decided to build the project as it seems quite ambitious and I'd like to see where they're at with it. I guess we will see what we will see.

      --
      No, no, you're not thinking; you're just being logical. --Niels Bohr
  • (Score: 1) by gawdonblue on Saturday March 14 2015, @11:07PM

    by gawdonblue (412) on Saturday March 14 2015, @11:07PM (#157894)

    I let Sirius answer that for itself:

    Me: "Surely you can't be Sirius?"
    Sirius: "I am Sirius. And don't call me Shirley."