Dealing with Poor Audio

Download as PDF

Introduction

It's in chopsticks?
No, I said it's in chapter six.

One of the biggest complaints we get from the clients is not Style Guide errors or sentence structure or even punctuation issues; it's missing and misheard words. Yep, you got that tight. And please be aware that if it is a real word, Word and Grammarly may look right fast it.

Unfortunately, as opposed to grammar, this issue is not something that we can "teach" you how to do properly. We can, however, offer you some tips and tricks.

A Word Regarding Poor Audio

We do understand that, at times, clients are going to load not so great audio. Due to the nature of our business and the wide range of clients, audio will not always be perfect.

We do reach out to the clients with our concerns regarding the audio quality, but sometimes it is just a small section of the file and/or the client is aware and still wants the file done to the best of our ability. They also understand that if they have lower quality audio that their files will contain more inaudible words.

At TranscribeMe, you always have the right to cancel out of any file for any reason. That said, the files still need to be completed and we need to try our best.

At transcription, we give a very high 30% of change to a file and this is one of the reasons. Your percent of change on files is not the only metric we use to determine your status. As a matter of fact, the transcribers we consistently see willing to take on those challenging files are considered for other projects that might require that extra effort and determination to get it right.  So please always put your best effort into those harder files and don't be scared about your percent.

If you are a transcriber and see a change made at QA that makes you think, "Oh my gosh. Why didn't I see that?" or conversely, "That's not what they said," don't lose heart. The QA had the whole picture. The speaker may have repeated the same word or phrase more clearly at another point in the file.

QAs, please keep in mind what it was like as a transcriber and are given only four minutes or less to determine the subject matter. If you look back on your work history, I'm certain you will be amazed at some of the mishears.

Some Tips, Tricks, and Good Advice

1

Make sure to work in a quiet area or use noise-canceling headphones (preferably ones that go over the ear). We have also found that using an audio enhancer like FxSound (formerly DFX) or the free Ears for Chrome extension can sometimes help with this problem.

2

The vast majority of people speak English very well (even the ones that have accents). So paying attention to normal grammar becomes a handy way to catch mishears. One thing that helps is to try to follow the conversation. People don't just say words, they say sentences, and the sentences together form stories and ideas and interactions. If you are able to catch the speakers' drift, you will be able to spot words that seem out of place.

3

Listening to the file again is imperative as it is often surprising how much we miss initially, especially once we've acclimated to the speakers' voices just to see if you can grab an extra word here or there. Then, read through the session without the audio. This will help you see and realize that some of the misheard words do not fit with the context and the file as a whole.

4

For those tricky spots, try both slowing down and speeding up the audio. Sounds weird, I know, but it works. Slowing down makes sense, but you'd be surprised how, by speeding up the audio, sometimes the tricky words will seem to pop out.

5

You can also try playing with your volume levels. It may seem counter-intuitive, but sometimes turning the volume down can actually help, particularly in cases of loud background noise. Specifically, try turning the volume down on the hub but up on your headphones for a clarifying boost.

6

When trying to capture crosstalk, I find it easier if you concentrate on one speaker's voice at a time.

7

And lastly, context, context, context. If the speakers are discussing a book, lecture, course syllabus, etc., they are most likely talking about chapter six and not chopsticks (see example above).

I do hope that you can start using a tip or two today to help improve your skills right away. Please do keep in mind, too, that experience does go a long way in this particular area. The more you practice, the better you will get! You may just find yourself tackling those tricky ones you previously avoided.

Audio Enhancers

FxSound (formerly DFX) Audio Enhancer: Many, many transcribers swear by this software for enhancing hard-to-hear audio. There is a free trial available; however, if you plan to transcribe long-term, you may consider purchasing the full version of FxSound. It has many additional features. Specifically, it has been recommended to go into the Presets menu, choose Select, and then (12) Vocals/Speech Booster.

Ears for Chrome Extension: Ears for Chrome is a free extension on your Chrome browser. Many people are happy with this option as well. Some even find it better than FxSound.

Some tips for Ears from your teammates:

• Play the audio (without trying to transcribe) with Ears on. Then pick each knob and move to the top/bottom/left/right and listen for what's being changed. Continue moving it until it turns the file into better quality.
• The bar that goes across the entire area and is currently at '0' moves up and down. It will increase or decrease the overall volume of the audio. If your audio has a awful background noise, static, hum, etc., try moving the bar down to lower the volume of the overall piece, then move the dots around until you increase the volume on the voices.
• To clarify some voices, fiddle with the 1280 and 2560 Hz knobs. Take them a little downwards and to the right. When it's noisy/scratchy, try fiddling with the first blue knob ( the one at 40 Hz). Take it downwards (around -10 to -15 dB) and towards the right (around 640-5120 Hz)
• To reduce bass/ to clarify heavy voices, move the first purple knob at 20 Hz downwards (-10 to -15 dB) and to the right (around 160-240 Hz).
• Once you've moved the levels to something that is helpful, in the text field underneath the graph named "Preset Name" type in a name for those changes. You can name it anything you like, and then click "Save Preset." For every file, you listen to and have changed these levels to fit, repeat this process with new names for each new preset. In future files, you can quickly click on the presets under the graph to not have to adjust the levels manually each time.

Boom 2: For Mac users (only), Boom 2 is another option.