Skip to main content

OCR vs Transcription

Well. Yesterday was a revelation. One of my neighbours has started work on their biography. Nothing unusual in that, however they decided to hand-write it (they aren't on the internet). Again, nothing that unusual there. Now, they gave it to me to see what I could do to publish it. I publish a magazine, so while that's the first time, it isn't much of a stretch.

Unfortunately we then hit the problem; how to convert the hand-written documents into something we can print.

Here is an example of the documents;


As you can see, this is going to be a challenge to convert, but as it has been a while since I did anything like this - I'd give it a go; technology moves on, after all.

The first thing I did was feed it through OCR.

This was the result;

: S\\pot ad ous
| J og born GLE a poorbetel bear
Dor ew Bar AAD NAS, Gms Won
‘+ ogn As ugh Ke wena ow ea
Dn RAMA sre,» WD Nex
Vay done to om Se ca eel
vig oes MA Ww PRL a, ReQue Dood
- \\0s eet DRA EBS BeoomoKes
: comaker and ALararer yh London , My
Yate Sah or’ mss we Woke Sec WY Ladoga
' ‘London Oe. Voregdh» Se Dog ‘oer Mm YS
Ow woes ev ocrkdked Whthe Cod ion lds
Yo Gu Rered Ram Nottina aur worked ag
a Viki welbe KOT Creee, ond
Hlackwe aso Ww London Troy Log
Roe PNR? Laas Lensnctar Ro ue
Bak A anQhore, WO adage lad o
RSW, be ees oe Crierans Row Ceoszy
amd Ra WO). HEV HK toss Hy

Now, that is clearly corrupted rubbish.

On the off chance, though, I decided to feed it into OpenAI ChatGPT and see what that managed to come out with. OCR yielded the same "garbage" result, but it had an intriguing offer at the end - would I like to see it transcribed. I said "Yes" and it returned this;

"I have a lot to say

I was born in 1948, a wonderful year for the start of 1948 NHS, & was born in bed on 25th August 1948, bottom and legs first, and am an only child. I have hated being an only child all my life. Whilst writing this, I am 76, soon to be 77 in 2025.

My mother died in 1998, my father about 1985. My mother worked as a seamstress for Norman Hartnell, the Queen Elizabeth dressmaker and designer in London. My late father was a baker for Fitzroy Lodge in London nr. Lambeth. He was born in 1913 and was evacuated in the Cotswolds. He suffered from asthma but worked as a fitter welder for Crosse and Blackwell, also in London, then was transferred to Kent, so had to commute from Kennington. He was lent a chauffeur. We always had a fresh turkey at Christmas from Crosse and Blackwell—lovely on Christmas Day."

I'm speechless - utterly speechless.

To say this has opened up a whole new avenue for achieving is a complete understatement.

Comments

Popular posts from this blog

Acornsoft ViewSheet User Guide (Second Edition) 1986

 This is a high-quality scan of the second edition of the ViewSheet manual published by Acornsoft in 1986. This is available on the Internet Archive here .

A New Lidl for Northstowe (Leaflet)

This was delivered to the surrounding residents. Fingers crossed it gets approved! Here's the leaflet on the Internet Archive.

The Manor Community College - Prospectus from 2001

Here is a 2001 prospectus from The Manor Community College (a secondary school in North Cambridge, United Kingdom, that has now closed and re-opened as an Academy). The prospectus is available from the Internet Archive here .