Speed of transcription

Since I had never really timed the speed I do English transcriptions, I did a little experiment today trying to transcribe lecture 5.8 for one of the classes that I’m taking. The result? It was shockingly time-consuming: I spent 5 whopping hours transcribing (plus 3 more hours syncing) a 19-minute video. That’s even slower than a Cantonese job I had done that I had timed. To be fair, the lecturer talked extremely fast, there were virtually no pauses, and it required some research (yes, you read it right — there were personal names, and the names were not in the video). But I didn’t expect to require 15 minutes to transcribe 1 minute of audio. So I took the SRT file, removed the time codes, and fed the result to my word counter. 4347 words. Divide that by 19, and I got an average of about 228 words per minute. The lecturer was speaking almost 4 words per second. So that’s why the transcription took me so long, I suppose. So how do people actually estimate how long a transcription would take? How many words are there in a “typical” audio minute? I’d really like some answers.