Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
Mozilla Input Devices Open Source

Mozilla Releases Open Source Speech Recognition Model, Massive Voice Dataset (mozilla.org) 58

Mozilla's VP of Technology Strategy, Sean White, writes: I'm excited to announce the initial release of Mozilla's open source speech recognition model that has an accuracy approaching what humans can perceive when listening to the same recordings... There are only a few commercial quality speech recognition services available, dominated by a small number of large companies. This reduces user choice and available features for startups, researchers or even larger companies that want to speech-enable their products and services. This is why we started DeepSpeech as an open source project.

Together with a community of likeminded developers, companies and researchers, we have applied sophisticated machine learning techniques and a variety of innovations to build a speech-to-text engine that has a word error rate of just 6.5% on LibriSpeech's test-clean dataset. vIn our initial release today, we have included pre-built packages for Python, NodeJS and a command-line binary that developers can use right away to experiment with speech recognition.

The announcement also touts the release of nearly 400,000 recordings -- downloadable by anyone -- as the first offering from Project Common Voice, "the world's second largest publicly available voice dataset." It launched in July "to make it easy for people to donate their voices to a publicly available database, and in doing so build a voice dataset that everyone can use to train new voice-enabled applications." And while they've started with English-language recordings, "we are working hard to ensure that Common Voice will support voice donations in multiple languages beginning in the first half of 2018."

"We at Mozilla believe technology should be open and accessible to all, and that includes voice... As the web expands beyond the 2D page, into the myriad ways where we connect to the Internet through new means like VR, AR, Speech, and languages, we'll continue our mission to ensure the Internet is a global public resource, open and accessible to all."

Mozilla Releases Open Source Speech Recognition Model, Massive Voice Dataset

Comments Filter:
  • by Anonymous Coward

    I had a problem with voice assistants because it was not being done inside my circle of trust (but closer to a sworn enemy).

    If I can run this on my home server, it will completely change the game.

    I could now actually implement Star-Trek-style home automation, if I needed it.
    "Computer, switch mood to" evening'." (campfire color scheme lights, shutters down, adjust screen warmth, play some relaxing music, mix me some nice drink [I'm building a drink mixer] and connect me to a tight slut)

    • My first question, is 6.5% error rate considered good? Surely it has uses, but it seems to be good for "things that aren't important enough to write down right now, but I want a half-ass transcription just in case."

      Doesn't seem to be quite what an assistance would be writing down, one would hope.

      • by Anonymous Coward

        Error rates for speak recognition varies greatly. 6.5% is very poor is it was trained explicitly on your voice, it is stellar performance if you are to interpret a random person with foreign accent.

        While the article is sparse on details, the most common benchmark for speech recognition accuracy is the "Switchboard conversational speech recognition benchmark". I am assume they are referring to this test. The Switchboard benchmark test is made up of 40 phone conversations from native english speaking persons.

    • Tight?
      Sounds like your willy isn't that large...
  • Disconnect Alexa, SIRI and Google from the network, and shut them down.

  • They really need to train it on this...
    https://www.youtube.com/watch?... [youtube.com]

In case of atomic attack, all work rules will be temporarily suspended.

Working...