Loading ...
Sorry, an error occurred while loading the content.

Speech recognition

Expand Messages
  • Michael W. Dean
    Anyone used it? I used to have Dragon Dictate and didn t like it, but that was on a Pentium I. Anyone know a good one that I can feed a WAV file into? Thank
    Message 1 of 6 , Aug 1, 2007
    • 0 Attachment
      Anyone used it? I used to have Dragon Dictate and didn't like it, but
      that was on a Pentium I.

      Anyone know a good one that I can feed a WAV file into?

      Thank you,
      MWD
      "Clone The Homeless"
      Michael W. Dean's podcast that remembers when sex was safe and music
      was dangerous. (Free, and no iPod is needed to listen.)
      http://www.clonethehomeless.com
    • Chris McConnell
      Hi Michael-- We program using SAPI, Microsoft s Speech API. We use the recognition engines that ship with Vista (Microsoft English v8.0), and with MS Office
      Message 2 of 6 , Aug 1, 2007
      • 0 Attachment
        Hi Michael--

        We program using SAPI, Microsoft's Speech API. We use the recognition
        engines that ship with Vista (Microsoft English v8.0), and with MS Office
        2003 (Microsoft English v6.1). These are high performance engines. Other
        leading engines are more-or-less comparable (see the ongoing debate in
        http://tech.groups.yahoo.com/group/ms-speech/).

        For desktop dictation (i.e., dictating a letter), modern speech recognition
        is quite good. With basic training, you can expect 90% accuracy or higher.
        Dictation implies that you will avoid background noise, use a good
        microphone, speak carefully, and insert punctuation.

        For command and control (telling your system what to do), modern speech
        recognition is excellent. For example, our Personal Audio Link product lets
        you "talk to your PC" literally, for instructing your PC to recite e-mail,
        play podcasts, etc. PAL hears commands via local microphone (16 kHz audio),
        or via cell phone (8 kHz audio). With brief training, recognition is nearly
        perfect. For me, it's faster and better than any smartphone or other
        handheld device.

        Now, when it comes to converting a wav files into text, things get dicey.
        Unless you have dictated (minimized background noise, used a good
        microphone, inserted punctuation, etc) when creating the wav file, you may
        be disappointed. With "normal" sound recordings, like interviews, speeches,
        etc, recognition accuracy could be only 50% or worse. The SR engine may not
        have the benefit of being properly trained, and it may encounter many
        unexpected sounds. Under these conditions, general speech recognition will
        be difficult.

        At the request of MS-Speech group members, we created a utility for just
        what you mentioned: wav-to-text conversion. It's imbedded in our Adondo PAL
        (Personal Audio Link), mentioned above. Wav-to-text is not commonly used.
        It's launched from PAL's Watch Screen menu. You can download and try PAL at
        no charge, so feel free to experiment away.

        There may be other wav-to-text utilities out in the marketplace now. As of
        12 months ago, the only ones that we could find did not use modern speech
        recognition engines. That's why members of the MS-Speech group asked Adondo
        to create one.

        I hope this gives you a sense of the current state in speech recognition.
        Overall, progress over the last 10 or so years has been quite impressive!

        Chris

        Chris McConnell
        Adondo Corporation
        http://www.Adondo.net
        Now Playing on your Cell Phone: Email, Podcasts, Traffic, and more...

        -----Original Message-----
        From: podcasters@yahoogroups.com [mailto:podcasters@yahoogroups.com] On
        Behalf Of Michael W. Dean
        Sent: Wednesday, August 01, 2007 3:57 AM
        To: podcasters@yahoogroups.com
        Subject: [podcasters] Speech recognition

        Anyone used it? I used to have Dragon Dictate and didn't like it, but
        that was on a Pentium I.

        Anyone know a good one that I can feed a WAV file into?

        Thank you,
        MWD
        "Clone The Homeless"
        Michael W. Dean's podcast that remembers when sex was safe and music
        was dangerous. (Free, and no iPod is needed to listen.)
        http://www.clonethehomeless.com





        Yahoo! Groups Links
      • Michael W. Dean
        Thank you, Chris! MWD Clone The Homeless Michael W. Dean s podcast that remembers when sex was safe and music was dangerous. (Free, and no iPod is needed to
        Message 3 of 6 , Aug 1, 2007
        • 0 Attachment
          Thank you, Chris!

          MWD
          "Clone The Homeless"
          Michael W. Dean's podcast that remembers when sex was safe and music
          was dangerous. (Free, and no iPod is needed to listen.)
          http://www.clonethehomeless.com
        • P.G. Holyfield
          I have used a roundabout non-tech method for converting voice recorded braindumps into text. I will play the audio on my computer, holding my computer mic near
          Message 4 of 6 , Aug 2, 2007
          • 0 Attachment
            I have used a roundabout non-tech method for converting voice recorded
            braindumps into text. I will play the audio on my computer, holding my
            computer mic near the speaker. Then Dragon Naturally Speaking converts
            the audio into text which I copy into Word. It converts around 75%
            correctly, depending on proper nouns, my enunciation of words, etc.

            --P.G. Holyfield

            http://pgholyfield.com/maah - "Murder at Avedon Hill" podcast novel

            --- In podcasters@yahoogroups.com, "Michael W. Dean" <kittyfeet70@...>
            wrote:
            >
            > Anyone used it? I used to have Dragon Dictate and didn't like it, but
            > that was on a Pentium I.
            >
            > Anyone know a good one that I can feed a WAV file into?
            >
            > Thank you,
            > MWD
            > "Clone The Homeless"
            > Michael W. Dean's podcast that remembers when sex was safe and music
            > was dangerous. (Free, and no iPod is needed to listen.)
            > http://www.clonethehomeless.com
            >
          • Bishop James I Feel God Brown
            Dear Michael, I am a Dragon Evangelist!!! I have used the old versions (think I started at 5) and the current version. I was not at all productive until 7 or
            Message 5 of 6 , Aug 2, 2007
            • 0 Attachment
              Dear Michael,

              I am a Dragon Evangelist!!!

              I have used the old versions (think I started at 5) and the current
              version. I was not at all productive until 7 or 8 but version 9 (9.5 on
              vista) is EXCELLLLLLEEEEEENNNNNNTTTTTT. Even though I speak with ebonics
              - It understands!!! Training now takss on 15 minutes MAX.

              If you want to transcribe from a recording, and you have a digital rig,
              spend the 15 minutes training with the same rig you will record with.
              It is not mandatory - but it will make you feel better.

              The new dragon will even insert anticipated punctuation, but if you talk
              like I write, you will need to insert and change some of it later.
              In Him,
              JMb <><


              Bishop James 'I Feel God' Brown
              The Internet's Favorite Pastor
              www.1000churches.org <http://www.1000churches.org/>
              www.ifghosting.com <http://www.ifghosting.com/>
              www.ifeelgod.org <http://www.ifeelgod.org/>
              www.youcanplaygospel.com <http://www.youcanplaygospel.com/>


              --- In podcasters@yahoogroups.com, "Michael W. Dean" <kittyfeet70@...>
              wrote:
              >
              > Anyone used it? I used to have Dragon Dictate and didn't like it, but
              > that was on a Pentium I.
              >
              > Anyone know a good one that I can feed a WAV file into?
              >
              > Thank you,
              > MWD
              > "Clone The Homeless"
              > Michael W. Dean's podcast that remembers when sex was safe and music
              > was dangerous. (Free, and no iPod is needed to listen.)
              > http://www.clonethehomeless.com
              >




              [Non-text portions of this message have been removed]
            • tom_a_sparks
              ... another way it could be done is by setting the input to your mixer line and playing the audio file on your computer
              Message 6 of 6 , Aug 2, 2007
              • 0 Attachment
                --- In podcasters@yahoogroups.com, "P.G. Holyfield" <pgholyfield@...>
                wrote:
                >
                > I have used a roundabout non-tech method for converting voice recorded
                > braindumps into text. I will play the audio on my computer, holding my
                > computer mic near the speaker. Then Dragon Naturally Speaking converts
                > the audio into text which I copy into Word. It converts around 75%
                > correctly, depending on proper nouns, my enunciation of words, etc.
                >
                another way it could be done is by setting the input to your mixer line
                and playing the audio file on your computer
                <snip>
              Your message has been successfully submitted and would be delivered to recipients shortly.