How to Create Transcripts Using Voice Recognition Software

One of the greatest features of voice recognition software is that it can be used to create transcripts--either in real time or of a recorded voice file. Though there are some limitations, for people who attend many meetings or make a living interviewing subjects, this can relieve a lot of the tedium of creating transcripts manually.

Things You'll Need

• Audio recording
• Microphone
• Word-processing program

Instructions

1. Purchase and install voice recognition software, using the installation instructions to set up the software. Make sure your computer meets the minimum system requirements for the program you choose.

2. Create a voice profile. When configuring the voice profile, remember this a profile for the person whose voice you want to transcribe, not your own voice. If you use your voice, the other person transcribed won't be understood by the software.

3. Determine whether your voice recognition software has a transcription mode. A transcription mode, unlike a speech-to-text mode, will understand that no corrections should be made, so it won't assume that any given word is right or wrong. Since it's a different function, using a transcription mode will not negatively affect the training of your voice recognition software.

4. Open a word-processing program and create a new document. This is the text document into which your transcript will be written.

5. Play the recording either into your microphone or via a cable in the microphone jack. Playing the sound directly into the microphone jack is recommended as it adds the least amount of noise.

6. Make necessary corrections manually. You may need to add sentence structure and punctuation to the document. In transcription mode, it's expected that corrections will be made as text is entered.

7. Verify that your transcripts are correct. Aside from any obvious mistakes, there can also be subtle mistakes, especially in noisy or quiet recordings. Play your recordings while reading the transcripts to be sure every word is correct.

Tips & Warnings

• If the person on your recording is not available to train the voice recognition software, it will be extremely inaccurate at first. It's best if you can get the person to train the voice recognition software as you're setting it up, especially if you will be using multiple recordings of his voice.

• The recording can't be played on the same computer as the voice recognition software unless your computer can "record" from the speaker output (some sound cards can do this, but most cannot). This is because most sound cards have "echo cancellation" to prevent feedback from the speakers.

• Don't play a recording repeatedly (as when making corrections). Doing so may negatively affect the training of your voice recognition software.