How to Transcribe an Audio or Video File With Bad Sound Quality
January 4, 2018
How to Get the Best Out of Dodgy Audio
I have worked as a transcriptionist for the past eight years and have learned quite a few handy hints along the way on how to get the best from a poor quality piece of audio.
Here’s just a few:
· Turn it up! I use Express Scribe transcription software and there are many options for tuning poor quality audio. The first I try is the channel toolbar. Dictation audio often includes a floor threshold reading. Using the channel bar on Express Scribe, you can alter the dB (volume) of two channels within the audio. This can really help to make a speaker’s voice stand out. For extremely quiet, or distant speakers, a better option may be Extra Volume Boost within the Special Audio Processes menu. This can be used if the volume needs more power than the main controls give you. Extra Volume Boost increases volume by 10%.
· Fine Tune it! For extremely fuzzy sound, there are other Special Audio functions in the Express Scribe software. For fuzzy or muddy sounds, you can ‘Reduce Background Noise’ or use the ‘High Pass Filter’. Reducing the background noise should be your first option if the speaker is audible but possibly driving in a car, sitting in a cafe, or rustling papers whilst he dictates. In this function, the Express Scribe distinguishes sound by volume, so if the volume of background noise is almost equal to that of the dictation, using the High Pass Filter may be a better option. The High Pass Filter distinguishes sounds by pitch so here, only sounds above 450Hz are kept in the audio. This is a highly valuable tool for clearing up ‘muddy’ audio.
· Skip it. When you are typing a transcript and can’t hear a word, it’s very tempting to keep pressing that foot pedal and listening to that same word again and again. The truth is by doing this, you are actually making your job harder by taking the word out of context. Try noting the unclear and then listening to the end of the sentence, or even the next few sentences. Chances are that word may be repeated, and if not, you still increase your chances of deciphering it by considering, ‘What word is actually missing from this sentence?’
· Give it Time. Ever noticed that transcription seems to get easier midway into the audio? Sometimes it takes a while to get used to a speaker’s accent or to pick up jargon. If you find you’re getting a lot of unclears during the first five minutes or so of your audio, don’t get hung up on it. Note your [unclears] and carry on typing. You’ll most likely find when you go back to it later on, you will have adjusted to your speaker’s accent, pitch and style and will be able to catch that word a lot easier. Also, listening to one word again and again can give you a type of mental block. It can be helpful just to take a break by finishing your typing and then going back with a clear mind.
· Share it. If you have someone to hand, get them to have a listen. Two heads is better than one and you’ll be surprised how effective this method is.
· Guess it. – Well, sort of. This tip tends to work well for industry specific jargon. If you think a word is a company name, product or industry specific term, it can be very worthwhile to simply type it phonetically into Google and see what comes up. We all know that if you use the Autocomplete function in Google, the search engine gives you suggestions of words. Watch those words as you type each letter and check if any seem right. If one does, click onto it and check the meaning to see if it makes sense within the context of your transcript. Alternatively, type your word, and see what spelling corrections Google comes up with. This method is brilliant for finding the correct spellings for medication when taking on medical work. Beware, however, that you do check the meaning, spelling, and relevance of the word, and also listen to it one more time in the audio, just to be sure.
So if you are offering transcription services, you can be sure that with these tips you can always get the best from any audio file.