The definitive guide to better transcription.
How to transcribe semi automatically using dictation software
While, automatic audio transcription software is still science fiction, thankfully, we’ve made a lot of progress.
In fact, there is a neat trick that leverages current voice recognition technology to transcribe nearly any type of spoken audio recording.
As we mention here, current technology only supports single speakers and can easily get confused when there’s background noise or interjections in the audio.
However, it works well for audio with a single speaker, when words are spoken consistently at regular conversational pace. If we can convert our multi-speaker, noisy audio into a clear single-speaker audio input, we will be in a better place.
And that’s where we use dictation:
- Load the audio/video into Transcribe and slow it down if the speaker(s) is (are) too fast.
- Connect your headphones to your computer and turn on Transcribe’s dictation feature by clicking on the green “Dictate” button.
- Now, repeat what you hear in your headphones into your microphone. You are now effectively converting your source audio into a single voice (your voice) and giving it as input to the voice recognition engine which is trained to recognize a single speaker.
Of course, the recognition can still be a little off, so be prepared to go over your dictation-transcription a second time to correct typos and add in punctuation and formatting.
When you first click on the “Dictate” button in Transcribe, Chrome will ask you permission to let Transcribe access your microphone. Make sure you click on “Allow”:
As a bonus, the dictation engine supports several languages including English, Spanish, Portuguese, Dutch, Chinese, French, Hebrew, Italian, Japanese, Korean and Russian. If you would like us to add support for another language please let us know.
As always, if you have any questions or if you think this article needs more clarification, let us know at firstname.lastname@example.orgTweet Follow @wreallystudios