Transcription Training


image
Windows
-->

Transcription Training teaches MacSpeech Scribe to recognize a voice and improve its vocabulary and accuracy of the Profile.

You can use the trimming markers to choose a portion of the file to transcribe, or you can leave these at the beginning and the end (as they appear by default) to transcribe the entire file.

Click the Transcribe button to begin transcription training. MacSpeech Scribe will begin analyzing the file and producing a transcription. The transcription appears in the upper portion of the Transcription Training window. Once MacSpeech Scribe has transcribed 15 seconds of audio, it stops to let you make corrections.

Check the transcription for any errors. If you see any corrections that need to be made (and it’s likely there will be some, since Scribe isn’t yet familiar with the voice and vocabulary used for this profile), click on the text that contains an error. The correction window appears.

Note: In the transcription results, punctuation marks are displayed as words. Do not change these words to symbols in the Correction window.

Make corrections as necessary to the transcribed text. If you are unsure, you can click the Play button to play back the portion of the recording currently being transcribed.

Once you are happy with the transcription of a section, click the Accept button. Or, alternatively, if there is a portion of the recording you want MacSpeech Scribe to ignore and not attempt to transcribe (extraneous sounds, for example), click the Ignore button. In either case, the correction window will move automatically to the next phrase.

Each section of the transcription you have accepted or ignored is changed to green or red in the Transcription Training dialog, letting you check your progress. Once you have either accepted or ignored all of the phrases in the first fifteen seconds of the transcription, you are ready to calibrate (let the program use the corrections you have just made to improve its transcribing ability).

Click the Calibrate button to continue. MacSpeech Scribe will be busy for a few seconds, analyzing the transcription corrections you’ve made and calibrating its next transcription. Once your corrections have been analyzed, the program will transcribe the first 90 seconds of the recording.

Repeat the correction process described above to make changes to the first 90 seconds of the transcription. Once you have made that set of corrections, the correction window disappears and the Train button appears.

Click the Train button to continue. MacSpeech Scribe will be busy for a few seconds, analyzing the transcription corrections you’ve made and further training its ability to accurately transcribe your file. Once your corrections have been analyzed, the program will transcribe the remainder of the file.

Please note that correcting and accepting a phrase may have to be performed more than once in order to obtain the desired result.


image
Transcription Vocabulary Training
image