There are different reasons that are affect transcription quality:

Heavy background noise

This is by far the biggest enemy of high-quality transcription. As far as possible, try indoor or in-studio recording, as noise editing is quite an arduous task. If you have background noise, firstly eliminate it.

Platforms like Audacity, Adobe Audition, and AVID have tools to help, but it’s not a sure fix. Audacity is a free open-source audio recording and editing application. 

• Open Audacityteam.org/download, click “File” and “Open” and locate the file you want to fix

• Select your “Room Tone” by dragging your mouse over a section where there is little or no audio sound

• Select the “Effect” menu and then select “Noise Reduction”

• Click “Get Noise Profile” and wait for the modal to disappear

• Click back to your timeline of audio to clear your selection

• Select the “Effect” menu again and select “Noise Reduction” again

• Click “Preview” to hear the default noise removal settings. You can select one of these defaults to see if they work

• If you still hear noise, then simply adjust the Sliders (see descriptions below) and then click “Preview” again

Keep the microphone close to the speaker

The closer the microphone is to the speaker, the better. Enunciating the words well will ensure better sound fidelity.

Skype and phone interviews

These typically have very low recording quality due to the compression used in these services. You may encounter lower accuracy with these recordings.

Heavy accents

Our models are trained in a lot of data, but heavy accents are one of the most difficult things to transcribe!