Automatic transcription of audio and video files is a very useful tool for journalists and those who often work with interviews in their jobs –market research institutes, for example. Whilst up to now, recorded conversations had to be typed out in tedious, hours-long bouts of manual work, now it’s done for you by artificial intelligence in the form of speech recognition. Recordings of spoken language are converted into text within a few minutes, and only minor edits and corrections are needed.
Speech recognition software now has a very high recognition rate, and the subsequent effort correcting text is incomparable to the effort of manually transcribing your audio recordings. But you can also cut down on this effort too as the sound quality of your audio file is a decisive factor in determining the accuracy of your automatic transcription. And this is something you can very simply improve.
It goes without saying that you shouldn’t exactly record an interview next to a four-lane motorway or a construction site. So, when planning where to meet, you should already be thinking about having access to a quiet room away from the open-plan office space and checking you have closed all the windows and doors. There are nevertheless sources of noise that you hardly notice yourself but that effect the recording quality; namely, electrical devices such as printers, air conditioners, refrigerators, and computers. It’s best to switch them off completely during the interview itself.
If you’re using a computer for the interview, don’t place the recording device directly next to it in order to avoid any disruptive noises from its built-in fan.
a. Everyone has one, everyone always has it with them, and it is very practical – the smartphone. In perfect environments, the smartphone may have good enough quality for recording interviews, but a built-in microphone quickly reaches its limits, especially if you are recording more than two speakers.
b. Dictaphones are typically equipped with high-quality microphones and definitely provide higher quality audio than smartphones.
c. If interview people a lot, it may be worth buying an external microphone to guarantee optimal results in your transcript later.
Lapel microphones, i.e., clip-on microphones, are suitable for this – small in size but decent in quality. They are usually clipped to your clothes, don’t get in the way, aren’t distracting and remain at the right distance from the speaker automatically. The sound qualities these microphones provide is excellent.
Alternatively, table-top microphones are also a good option. There are different types – the traditional omnidirectional microphone, which picks up sounds from all directions; or the so-called directional microphone, which picks up sound in front of it better than ambient noise.
Whichever you choose is, lastly, a budget question.
d. If you want to work perfectly, you should use a pop screen or windscreen on the microphone – it helps to reduce the pop caused by plosive sounds like ‘p’ and ‘b’. And it is best to position the microphone, so it points at the speaker’s mouth and not at their chest or chin.
You already know which questions you will use to moderate the conversation. You do not yet know what the person you are interviewing is going to answer, though, and that is what you want to find out and record. So get as close as possible to the speaker with your microphone or recording device – and as close as recommended with a microphone. The distance between the microphone and the speaker is a decisive factor in determining the recording’s quality and it helps to record only the speaker’s voice while minimising background noise.
It is definitely worth optimising the audio quality of your recordings in advance. High-quality audio makes a considerable difference in your transcription software’s recognition rate – you get a transcript almost entirely free of errors, save yourself the effort of correcting it and can get on with your actual work straight away.
Curious? Take advantage of the opportunity to test GoSpeech for free and let our transcription software transcribe your audio and video files.