Transcription is the process of writing down what is said in an audio recording. It seems like a simple secretarial skill, whose main requirement is an ability to spell.
That is until you try it
When you do, you find that ‘writing down’ is not the main issue. Before you get there you have the ‘what is said’ part. Even in a clear recording, that can be surprisingly difficult.
Getting the overall meaning of the conversation is easy, but transcription requires rendering each individual word and phrase. Even in clear speech, that is far harder, taking an amazing amount of checking and re-checking. And the chance of missing errors is very high.
It’s a bit like proof-reading, but worse
Ever read a book and discovered a typo? Ever thought about how many times that page has been checked and double checked and triple checked by how many people? Mistakes are easy to miss in print, and far more so in speech.
I nveer miss typos, do you?
So have respect for people who work as transcriptionists!
Transcribing indistinct audio is way harder
When we start dealing with indistinct recordings, things get far harder. It takes many many hours of careful, patient listening, checking and re-checking to get even a basic transcript.
And then comes the really important issue – how can you (or anyone) know your transcript is right?
This is where the law makes its big mistake, by developing its process on the basis common knowledge.
Common knowledge suggests that to check a transcript, you follow the written version against the audio – and note down any mistakes.
Seems simple – until you try it
Indistinct audio goes past your ears very fast – much faster than you can read a transcript (especially the kind of transcript typically produced by police). And covert recordings are often very lengthy, hours or even days of material.
Following the written version means constant rewinding, trying to find your place. It’s not just tedious. Poor quality audio is very unpleasant to listen to. It is like listening to an off-station radio, but worse. It’s not just crackly. Your ears are assaulted by sudden loud clicks, whines and whistles.
Hardly surprising if some lawyers are willing to outsource this task to the jury, right?
But there is a worse problem. Even if lawyers take on the responsibility of checking their opponents’ transcripts diligently, they are unlikely to notice all the errors.
That’s because the transcript itself can influence your perception in ways you don’t even notice. You feel you are ‘hearing with your own ears’, but you are really being primed by the transcript.
And once you feel you have heard words ‘with your own ears’, it is remarkably difficult to un-hear those words, even when you know intellectually that they must be wrong.
There’s heaps of information and additional demonstrations on this topic, with in depth discussion of the injustice it has caused in real cases, over at Forensic Transcription Australia.
There’s also loads of information in RTS1: Unlearning to help you understand what’s wrong with the apparent truism that ‘you hear what is there to be heard’.
Or you might choose to continue to the next topic in this Forensic Phonetics module, that tackles the commonly asked question: Why don’t they just enhance indistinct audio to make it clearer.
Whichever one you choose – enjoy!