Tips for Improving your Transcriptions!

Welcome to the Biweekly Friendly Tips post! 

Written by fellow CrowdSurfers, these posts are published every Monday and will list the previous weeks most common errors and help with how to avoid making those errors in future tasks!

Monday, December 12, 2016

Hello, CrowdSurfers!

There was some very poor audio as well as some heavy accents coming through the queues last week which resulted in a lot of [INAUDIBLE] and [UNKNOWN] tags in FTR. It’s very important that we replace these tags with the correct transcript whenever possible. The beauty of an FTR, as opposed to just a short transcription snippet, is that we get to listen to an entire, several minute piece, thus getting a pretty good idea of the subject matter and of any speech pattern peculiarities the speaker may have. Also, a quiet working environment and a pair of headphones or ear buds are essential. I don’t have a high-tech headset or any audio enhancement software, just a very basic pair of headphones, but just with and paying close attention to the material, I was able to replace well over half of [INAUDIBLE] and [UNKNOWN] tags I encountered in QC and QC Reject this past week.

There were also a remarkable number of instances where the Clean Verbatim guidelines were not followed. Repeated words and fillers such as um, like, and you know were included. Remember, the default guideline for all CrowdSurf transcription is to use Clean Verbatim. If a piece should be True Verbatim, there will be a pop-up before you begin and/or special instructions to that effect in the sidebar notes.

Lastly, Speaker Change errors were a continuous problem, both having them where they didn’t belong and not having them where they were needed. Remember, we need a speaker change indicator immediately before each new human speaker but not before sound tags indicating blank audio or non-human sounds. Also, when working with poor audio, it may not always be possible to determine what’s being said, but by paying attention to the tone of the voice, we can very often determine when a different person starts speaking. So [INAUDIBLE] >> [INAUDIBLE] is a perfectly acceptable transcription and can often be helpful to the next person to work on the piece.


Wishing us all Full Queues and Clear Audio!

Peace, Love, Transcription!

For more help with Full Text Review tasks, check out the links below: