In article <slrn7n03s8.fi.aiyar at ebv.oncology.wisc.edu>
aiyar at ebv.oncology.wisc.edu (Ashok Aiyar) wrote:
: A member of our group is currently entering a large number of
: short sequences into a database. Are there any applications
: that do speech to text conversion so that she can enter the
: sequences verbally?
: I have AppleScripts setup that enter the characters A, G, C,
: and T into a Microsoft Word document when she says "apple",
: "go", "cat" and "to", using Apple PlainTalk Speech recognition.
: But this is very slow -- it takes about 10 - 30 seconds per
: character. It is therefore currently faster to type the
: sequences in.
: I would love suggestions for alternatives, or ways to speed
: up the entry process.
Other than OCR, there is no faster way than typing, IMO. I can
certainly type a sequence faster than anyone can read it out. Other
ways, like getting someone else to read out the sequence while you type
it in, are surprisingly common but dreadfully slow and error-prone. The
thing to do is teach four of your favourite fingers where the A, C, G
and T keys are, then type into a file without looking at the keyboard
or the screen, i.e. only looking at the input sequence. It's quicker to
learn to do this than you might think. Then (get someone else to?) type
in the whole damn thing again, format the files and compare
electronically (diff on Unix). This picks up the typos. I can reliably
do more than 200 characters a minute (using my left hand for A and C
and right hand for G and T), with an error about every 1000. These
days it's hardly ever necessary to type in long sequences, but I well
remember typing in the pBR322 sequence when it first came out in, oh,
the Anglo-Saxon Chronicle I think it was.
Chris Boyd | from (but not \ MRC Human Genetics Unit
Christopher.Boyd at hgu.mrc.ac.uk | on behalf of) / Crewe Rd, Edinburgh
http://www.hgu.mrc.ac.uk/Users/Christopher.Boyd EH4 2XU, SCOTLAND