These past two days have been a revelation – although not a good one. and I’ve been left with just the embers of my project and on my knees, eyes streaming, trying to blow some life into them. The bright new day is yet to dawn.
Just to recap, I’ve been designing and now testing a new workflow for my botanical recording – attempting to make it faster, more accurate and less of a square peg when it comes to uploading into the round holed national databases; but rather like the time I rebuilt the engine of our Morris 1000 and left out the rotor in the oil pump, the flaws soon made themselves apparent when I road tested my bright new idea down here in Cornwall.
There’s a lane outside the campsite which begins as the metalled road down to a large house, presumably once part of a home farm at the centre of the big estate. It then continues as a footpath above what’s now a hay meadow for a couple of hundred yards, across a stile into a patch of scrubby woodland for a little way until it joins a very pretty deep sided sunken track which leads to the beach via another track heavily decorated with dog poo bags making the exploration of its botanical residents a bit hazardous. The fact that it’s Madame’s favourite walk makes it easier to smuggle in a bit of road resting. Sometimes. Although I’ve never formally surveyed the plants along with its several micro-environments, it’s the perfect place to learn. So here it is; the method …..
Step one attach the microphone and receiver ready to record all the details for later, and when the plant is spotted, record them in one continuous packet of data – Problem – I talk too much and instead of recording just the details, the recording features too many oohs aahs and miscellaneous grunts and strange noises. Worse still comes later when I load the recording into the transcription software.
Step two – take a quick photo on Flora Incognita to get a reference picture with a provisional name. Problem – no internet signal. But I do get a time and unreliable location so as to link the data together later
Step three – take any measurements that I might need later and note them for the audio
Step four – take reference photos on the Olympus camera which can do macro and has better resolution. Problem – stiff winds make focusing on grasses a complete nightmare. I either need to take a small piece of board as a background or uproot specimens to photograph later – not ideal and sometimes illegal.
Step four– take GPS reading and photograph it on the phone or note it. This ties it all together. Problem – remember to figure out how to reset the GPS to gather the current location and write it down (data separation hazard) or record it. Need to write a short script and stick to it.
Step five – record all the data; names, grid reference and measurements etc in a standardised form that AI can understand.
Step six– back at the ranch, run the audio through the transcription software and edit if necessary. This is where the system totally collapsed because I was using the free Google Record app which handles ordinary conversation pretty well, but went to bed with a headache at the sound of plant names and just fainted at Latin binomials. I tried to edit the result but it was terribly time consuming so I recorded a page from John Wright’s “Grass Lands” to give it another chance. This was a great laugh, with Bird’s foot Trefoil being rendered as Bird’s foot truffle oil, and the blameless little Eyebright – “Euphrasia nemorosa, the commonest form” unaccountably branded as the communist form. Obviously this level of inaccuracy is beyond the editing powers of any AI programme. The plan was to feed these pre-digested dollops of data into Notebook LM and ask it to sort them into a spreadsheet in correct order to pass through the pearly gates of head office; every single entry checked against the great canonical parchments.
Step seven – this was meant to be the bit where I closed the reference books and where I modestly admitted that the new workflow was a triumph, instead of which there was much wailing and gnashing of teeth. When I remonstrated with Google Gemini it almost cracked a robotic joke and said the bird’s foot truffle oil sounded like the salad dressing from hell! (now that is scary – it’s cracking jokes already).
However – I’m an eternal optimist – there are some tweaks I can see which fall short of carrying a mobile router around with an aerial tied to my bandana. It’s helped me see how to improve my photographs to get better results – especially with grasses and I’ve learned a lot about identifying grasses – which was one of my targets. I’m pleased to say that I’ve added five more grass species to my total. We also met a Swedish woman and talked about the yellow flowered strawberries that grow down the lane. We thought they were probably poisonous but she said they used to eat them in Sweden but they were insipid and not that sweet. She doesn’t eat them here because of the dog poo that gets everywhere. After a few days of wind and rain the sun has now come out and Madame is cooking a dish of ratatouille as a punishment for my distraction. Over the twenty years we’ve been coming here the village has become more and more gentrified. The local spar shop now sells curated wines, amalfi lemons and “pain de campagne” and an ice cream next door sells for £4.50 a scoop. Here are a few pictures from the entirely un-gentrified lane. They’re all weeds, but nonetheless rather charming in their blunt honesty. It’s a very under-recorded little paradise and one day I’ll bite the bullet and do a proper survey.
