Camtasia (Windows): How to configure Microsoft Speech Engine for Speech-to-Text

Problem

How do I use the speech to text feature in Camtasia.

Solution

Camtasia has a feature called Speech-To-Text which utilizes Microsoft speech engine to convert the audio in your presentation into Captions. The following is a summary of how to correctly configure the engine to get the optimum quality.

Installation

Windows 7, 8, and 10

Microsoft speech engine is part of the OS so speech engine install is not needed. After you install Camtasia, the speech recognition features will be ready to use.

Available languages: 

  • US English
  • UK English
  • German
  • French
  • Spanish
  • Japanese
  • Traditional Chinese
  • Simplified Chinese

 

Training Your Computer & Configuring the Microphone

For more accurate Speech-to-Text transcription of your audio recordings, we recommend you go through the Voice Training tutorials provided with the Speech Recognition software.

You are encouraged to train your computer and configure the microphone that will be used for speech dictation. Once you spend some time (4 hours, as suggested) training your computer, you don’t have to train it again.  You may export then import the profile to reuse the training info on different login or different computers. Users can have more than one profiles for each login. This also makes it possible that you transcribe audio files recorded by someone else on your computer, as long as they also send you their profile files. If audio is composed by more than one persons’ voices, you can’t get too much benefit by just using one profile - multiple profiles cannot be used concurrently.

Windows 7/Vista speech profiles can be managed with this tool:

http://www.microsoft.com/downloads/en/details.aspx?displaylang=en&FamilyID=1d60a5a6-85d4-4db2-a581-a41f66561a7d

 

Tips to Improve the Accuracy of the Speech Engine

  • Use the best speech recognizer you could get. For example, on XP, you may install Speech Recognizer 6.1 instead of default public domain version Speech Recognizer 5.1
  • Accuracy is improved by training and audio quality. Best accuracy requires 4-5 hours of training. The more you train your computer, the better result you could get. You may click the Train profile button in the first tab of "Speech properties..." window to train your computer
  • Custom words can be added to a user’s dictionary by telling the system the text word and speaking the word (e.g. you can explicitly tell the system to recognize how you speak the word “Camtasia”). You may access menu Add/remove words... to do this.
  • There are no acoustic models and audio quality settings for speech engine, however, on XP machine, you may set the recognition quality vs. recognition speech by clicking the Settings button in the first tab of "Speech properties..." Window.
  • Use a decent quality microphone and configure the microphone properly. You may click the Configure microphone button in the first tab of "Speech properties..." window to configure the microphone.
  • Use the proper training profile to do the speech recognition.
  • Record or dictate your voice in a quiet environment and use your normal speed to speak.
  • Choose a speech recognizer that best matches your accent (e.g. US vs. UK for English) in "Speech properties..." window.

You may also install MS language packs to obtain the speech engines in other languages.

Useful Links

Microsoft Speech API

 

Was this article helpful?
0 out of 1 found this helpful