Jump to content


Need Step by Step directions for speech to text


  • Please log in to reply
4 replies to this topic

#1 mkg

mkg

    Casual

  • Members
  • PipPip
  • 11 posts

Posted 29 December 2010 - 06:16 PM

I've been using Express Scribe for manual transcription, but want to use it for speech to text. I see in different areas, snippets of how to get this to work, but would like to have step-by-step instructions.

I'm using Express Scribe 4.23.

If I understand it correctly, you must do the following:

1. Have a speech profile loaded on the machine. (I recorded one using Voice Recognizer.)
2. Before you load a dictation file into Express Scribe, you must click Settings>Other Settings from the main toolbar, and click the Other tab.
3. Select the Use speech recognition engine checkbox, and select the engine (in my case, it was MS English (U.S.) v6.1 Recognizer).
4. Click the User setup... button. On the dialog box, I selected user ID zero (0) since my dictation file was a .wav file and the speaker profile was created in something other than Express Dictate; therefore no user ID was automatically assigned.
4. Click OK. The Express Scribe main page reopens.
5. Now load the dictation file that you want to have converted to text. After it is loaded, it is displayed on the Express Scribe main page.

This is where all available information ends.  Here are my questions:

o Do you double-click the dictation file after it has been loaded and let it run?
o How do you know that transcription is happening? I know that I read that it takes a long time to convert speech to text, but my recording was only 18 minutes long and I've been waiting for hours.
o Where is the area of the window where the text is supposed to display?
o How do you save whatever has been converted to text?

Thanks for any assistance.  I hope that your answers can help other users as well.

#2 nchdl

nchdl

    Guru

  • Support
  • PipPipPipPipPipPip
  • 452 posts

Posted 29 December 2010 - 09:28 PM

View Postmkg, on 29 December 2010 - 06:16 PM, said:

Do you double-click the dictation file after it has been loaded and let it run?
The speech-to-text conversion should start automatically when the dictation finishes loading. There's no need to click or double-click.

Quote

How do you know that transcription is happening?
Unfortunately, Scribe doesn't give any visual indication that speech-to-text is working in the background.

Quote

my recording was only 18 minutes long and I've been waiting for hours.
That's obviously too long. Something must have gone wrong. v4.23 is very old though, and there have been changes to the speech-to-text code since then. Can you try a more recent version?

Quote

Where is the area of the window where the text is supposed to display?
Speech-to-text output is pasted into the Notes window.

Quote

How do you save whatever has been converted to text?
I'm not sure about your old version, but recent versions have a File -> "Export notes" menu item. Or you could just copy and paste the text into a word processor document.

#3 mkg

mkg

    Casual

  • Members
  • PipPip
  • 11 posts

Posted 30 December 2010 - 05:13 PM

Thank you so much for responding.  All I needed to do was to upgrade to the latest version of Express Scribe and watch the magic happen.  I was able to get the text in a very short time. However, it was all garbage, which tells me that more training is required for the speech profile.

Speaking of speech profile, is there an easy way to import a speech profile into Express Scribe?

#4 nchdl

nchdl

    Guru

  • Support
  • PipPipPipPipPipPip
  • 452 posts

Posted 03 January 2011 - 09:14 PM

View Postmkg, on 30 December 2010 - 05:13 PM, said:

Speaking of speech profile, is there an easy way to import a speech profile into Express Scribe?
You can't import a profile directly into Scribe - you need to import it using the tool relevant to your speech recognition engine. e.g. For Windows built-in engine, try the instructions here.

#5 mkg

mkg

    Casual

  • Members
  • PipPip
  • 11 posts

Posted 04 January 2011 - 01:07 PM

That was exactly what I needed. Thanks so much for sharing.