Posts tagged ‘speaker identity’

Successful Audio Transcription

There are 2 main operational difficulties with the transcription of audio recordings of conferences and meetings – poor sound quality and no idea of who is speaking.
 
Poor sound quality comes from the use of voice audio recorders (that are primarily designed for close-to-mouth dictation) being set to RECORD and then being positioned in the middle of the table. The tiny microphone in these devices was never designed to deliver good quality audio under those conditions. Of course, sound quality degrades exponentially with the distance from device to speakers, environmental noise, table banging, paper shuffling etc
 
Not to mention the very difficult task of assigning a name to each and every conversational comment. So why not add speaker identity so that the typist does not have to continually REWIND in order to try and figure out who is speaking?
 
For these very reasons, many transcription companies work by the rule that it takes 6 hours to transcribe 1 hour of meeting audio of average quality – more if they want the transcript time-coded.
 
This of course results in 2 further economic problems:
 
– wasting the client’s time confirming who said what; and

– perhaps even wasting a lawyer’s time if they are the intermediary – and lawyers are expensive people to interrupt!
 
So, surely it would make sense to improve the capture of the original audio and provide a mechanism for speaker identity?
 
This would result in a win-win-win situation for transcriptionist,  lawyer and client.
 
The typist would be able to transcribe more quickly and more comfortably resulting in less fatigue – a common complaint with conference transcription work.
 
The lawyer would get a more complete transcript more quickly.
 
The client would get a faster, more complete service.
 
Everybody wins…

Technorati tags:

September 22, 2007 at 1:52 pm 1 comment


Categories

Feeds