Virtuadmin

When reading articles about marketing, interview techniques and the like, I do get concerned that businesses may be investing money in automated transcription services that do not provide them with the results of their expectations.

What do I mean by this?

Marketers, for example, record their interviews. Recording an interview allows them to keep track of what the respondent/s say, and not relying on note-taking. They can also listen back to the interview afterwards, whenever they need to.

If listening to the interview is not convenient for you each time, you can simply use an automated transcription service and have your recorded interviews converted into text. Reading is quicker if looking for something specific, rather than having to keep playing, rewinding and replaying again trying to find the specific place in the audio. This is all time-consuming.

It is the expectation of the layperson using automatic speech-to-text transcription services, that they will receive a decent transcript, not that isn’t error-free but certainly a lot more accurate than what most are expecting and receive, in reality. Automation does seem on the face of it to be one of the most efficient ways of getting back the results of your interviews whenever necessary. If the audio is clear, you’ll have your transcript in minutes.

But… what’s the catch?

Here’s the catch that I am about to share with you, the one secret that no one will tell you. Not the automated transcription services certainly. However, I will.

Below is a screenshot of an automated transcription services provider.

As you can see the process is very easy with three simple steps. However, the catch is in step 2. There, I’ve told you now! Yes, the catch is they’ve covered themselves that quote, “the benefits of our service, the accuracy can’t be 100%. But this is not a problem, you can easily edit your transcripts by your own.” Easily edit your own…?

Have you ever opened Windows Media Player and listened to an audio recording, either an audio interview, a podcast, music even. When listening to an interview, or podcast, for example, try rewinding just a few seconds if you misheard something. There is the problem.

Typing will always be quicker

From a manual transcriptionist’s point of view, and this I can speak of experience, it is very, very difficult to rewind manually just a few seconds to catch a couple of misheard words.

Any audio player will play audio, yes, automatically rewind five or ten seconds, no. It is not as easy or quick to correct an automated transcript, not without at the very least transcription software to play, rewind a few seconds and replay again.

Trust me, as a manual transcriptionist who recently edited an automatically created transcript from an audio file for a regular client, typing will always be quicker than editing!

The issues I have found with automated transcripts that repeatedly arise to date are the following:

1. When someone takes a pause where there should be a comma, automation inserts a period. In other words a full stop.
2. Therefore, when you are correcting full stops to commas, you’re also then having to normalise any capital letters.
3. For interviews, where there is an interviewer and respondent, which is usually the case the automation cannot intelligently define when the speaker changes.
4. Therefore, entering to create a new line when the speaker does change, tabbing across to separate interviewer and respondent identifiers, takes time.
5. Another annoyance is semi-colons for no reason whatsoever!