• Post Reply Bookmark Topic Watch Topic
  • New Topic
permaculture forums growies critters building homesteading energy monies kitchen purity ungarbage community wilderness fiber arts art permaculture artisans regional education experiences global resources the cider press projects digital market permies.com private forums all forums
this forum made possible by our volunteer staff, including ...
master stewards:
  • Nicole Alderman
  • r ranson
  • Anne Miller
  • paul wheaton
stewards:
  • Mike Jay
  • Jocelyn Campbell
  • Miles Flansburg
garden masters:
  • Dan Boone
  • Dave Burton
  • Steve Thorn
  • Greg Martin
gardeners:
  • Shawn Klassen-Koop
  • Pearl Sutton
  • Mike Barkley

Permaculture Planning Software

 
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I am starting some permaculture stuff (at least one year now, maybe two).  A friend asked, what software do you use for planning?  Sotware?  Sorry, I was just planning to keep it all in my head.

But, to do this with software (and a database and ...) makes sense.

Some seedlings I bought this last year, might be considered for harvest in 50 years or so.  This coming year, I want to set up a plot with a few different species, among which will be black locust.  And some of these trees I want to produce flowers for bees, some I want to use for growing fence posts, and some I want to use for growing larger dimensioned wood.  If I am lucky, maybe I can harvest fence posts on a 3 year schedule?

But permaculture is about having all kinds of things going on at the same time.  Few of which are on the same schedule.

I am interested in developing robots to harvest fruit and do maintenance on trees.

And then I have interests that have nothing to do with the aspect of growing things (plants or animals) on the "farm".

So maybe this kind of software is useful, possibly even necessary?

I run Linux, so  anything that is M$ or Apple only isn't of any use to me.  I am going to try LibrePlan, which is Java (and some other stuff).

In reading about Meltdown/Sceptre, JIT comes up a bit.  JIT gets mentioned a lot with Java, and so I wonder if this is going to have an affect on my choice in choosing software for planning?  One place where JIT cameup, is the Berkeley Packet Filter, and a particular instance of a Debian kernel.

=====

The last 2 paragraphs probably mean nothing to most people on Permies.com.  I apologise; don't worry about it.  The lead of Permies, seems to have more than a casual knowledge of Java.

 
Posts: 358
Location: Abkhazia · temperate subtropical climate
23
cat forest garden solar trees wood heat woodworking
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I have used a plain old text editor, a function plotter, inkscape and blender.

But mostly I use the Paper software with the pencil tool.
It is faster, has less bugs and art support. (I did learn technical drawing in school.)
I just can't be creative when working on a computer, so everything need to be done on paper first.

---
I would not worry too much about the latest computer security news. It just gets the perceived security level a bit closer to the real security (none, really).
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I took drafting in engineering school, but I am horrible at visualizing things.  I still have the tools.  Amazing a cardboard box can last that long.

I've had to follow security news probably since the mid 1980s.  It is part of my life now.

The new kernel on the machine I downloaded LibrePlan to, has Meltdown/Sceptre patches.  Using the mouse on the screen has stutter.  I expect once I figure out how to compile and run t his package, I will find it will stutter too.  Hopefully newer kernels won't hit AMD CPUs so hard.
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
If a person is working inside a factory, project management probably can run the way most of this software seems to push things.  If we look at civil engineering jobs exposed to the weather, we can see where things change.  At the outset, perhaps they design the project progression for perfect weather, and then have to insert delays as nature imposes them on the project.

In things like farming, everything is tied to the weather.

If I have fruit trees, I know that at some point in early spring (late winter), I need to try and make time to prune the trees in preparation for the coming year.  For any particular tree, I may know what I did the last time.  Did I get done everything that needed to be done?  Did I get everything done that I wanted to do?  Since then, I've observed the tree.  Are there other things I want to do?  Perhaps a day arrives, and it looks like "today is the day to prune tree A".  We would like to see a summary of what was done, what needs to be done, what is desirable.  And so we prune the tree.  This starts a clock of sort on the new project (another production season).  We may be able to guess that N days after pruning branches, that we may have to return to the tree to thin out blossoms or early fruit.  And so on.

If we work in the garden, planting seeds or planting seedlings started indoors earlier, we may at first guess know that we need to return on some kind of schedule to look for maintenance reasons and we need to harvest at some point.  Those growing day predictions are rough.  What is better is to known the base temperature for the plant in question, and then track degree days.  Are the degree days to harvest time related to other variables (like precipitation)?

We monitor weather stations on the property and in the vicinity.  We may have seen that we had strong winds the night before, and so a new thing TODO today has popped up: look for wind damage on trees (possibly even some trees are more prone to damage than others).

All of this shows that managing our projects involves a lot of other knowledge.  Which probably goes back to the original design stage (which may be ad hoc).  A couple of years ago I ran across some data on raising full size cattle near where I live.  It involved how much water was needed to grow feed, and how much water was drunk by the animals.  There are weather and vegetation (shade) components in both of those kinds of data.  In any event, using predicted rainfall (name, same as last year) I could estimate how many cows I could raise on my land based on how much water I receive.

Similar data should be available for other things, like planting trees.  If I am going to plant tree X, you should use a spacing which depends on your average precipitation and temperature.  The tree will require Y amount of water per year.  At a certain age, the tree reduces evaporation of water by Z due to shading and wind reduction.  And so on.  Maybe overplanting is always necessary, as catastrophes can happen and require the removal of a plant.

Not just plants.  Maybe we are planning to have chickens till the soil.  Bird flu comes along and all our chickens need to be destroyed.  So we need to adjust and adapt the plan.

Anyone seen this magic software?  

 
Sebastian Köln
Posts: 358
Location: Abkhazia · temperate subtropical climate
23
cat forest garden solar trees wood heat woodworking
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I would call this "software" people. People who have taken responsibility for a specific part. Like caring for the apple trees, a part of the garden where they can know every plant, a flock of animals, or predicting the weather.
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I was afraid of that.

You see, I am one of those people.  I started with computers in 1977.

If the only place to find this data, is to buy books, I won't get to it any time soon.  I have no budget for buying data.

In terms of making something useful, I am more of the type to find how to use the data.  Designing GUIs (or other user interfaces) is not something I am good at.
 
Sebastian Köln
Posts: 358
Location: Abkhazia · temperate subtropical climate
23
cat forest garden solar trees wood heat woodworking
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Well.. the problem with data is that you either don't have it, it is too unspecific, or you have to collect it yourself. (exaggerated a bit)

To me, a computer is great for working with definitive data, inputting a set of fixed values and seeing the result.
But in reality, definitive values are not often found. Everything has some uncertainty and when multiple sources are combined, this often increases.
Working with that kind of "spoiled data" is hard to get right in algorithms, but for my brain it is easy. Almost like it was build to process uncertain knowledge. (And I am sure that works for others too, once one stops thinking mechanically.)

The amount of data my brain can handle at once, may not as much as my laptops memory holds, but I don't have to translate it. I can easily keep a group of buildings up into the detail in my head. Displaying this in a CAD software will be a challenge, not speaking of navigating it! With landscapes, one would have to collect ground samples of the whole area to get usable data for a computer. And that probably every few days to track water and nutrients. Where as as a human I can look at the plants, the color of the grass the surroundings, put my hand into the soil and smell it. And my brain will automatically compare all that information against all other similar information seen before, and interpolate as necessary.

Why would I try to replicate this in a machine that needs to be programmed and does not reproduce itself?
 
Posts: 136
11
  • Likes 2
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I imagine you have already thought of this - a calendar which will send you notifications on dates you specify would be useful. That way you can plant something then set yourself a reminder to check on it at some significant date in the future.

The problem of course is that I imagine you would not want to use something like Google Calendar, as it requires a web connection and also is something you can't rely on 100% as it's a third-party service.

You've now given me an idea to make an open-source offline calendar based entirely on open web standards
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
A friend pointed me at something, which might be sort of based on a calendar.  I don't think it was in Perl?    Ruby?  There was a lot of stuff that was Java, and I thought this other thing was different.  I don't think it was PHP.

But, I got a new router that I need to set up.  I compiled OpenWRT for it (sort of), including a bunch of the mesh networking stuff.  So maybe that works with the calendar sitting on the server?

I have an Android phone, which will no longer receive updates to the OS, so maybe I can use that to talk to it?

I suppose that Matt hates working in Perl.
 
Matt Coston
Posts: 136
11
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator

Gordon Haverland wrote:I suppose that Matt hates working in Perl.


I'm completely ignorant of Perl. I only know the basics of PHP and JavaScript. I was thinking of something that would run on any modern computer without modification, regardless of OS, and I wasn't thinking of being able to share it over a server. If it was absolutely necessary I suppose I would use a raspberry pi with an sql database.
 
Sebastian Köln
Posts: 358
Location: Abkhazia · temperate subtropical climate
23
cat forest garden solar trees wood heat woodworking
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Why do you want to write new software?
Open source calendar software exists. It typicalls supports uses the iCal standard.
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Practice?

I looked at the Wikipedia iCal page, I don't think it will work.

Once upon a time (when I was C-MU), one of Brian Kernighan or Dennis Ritchie was there and gave a talk.  Little languages (such as tbl, eqn, ...) was part of this, there were other topics.  I like iCal in that it is an ASCII conversation.  The first thing that jumped out at me, is that iCal doesn't work with lunar calendars, such as are used in Israel and Saudi Arabia.  (What about the Moon, Mars, ...?)

I think this is more complex.

Let's say I want to do some three sisters (I don't think the fourth sister grows here?) planting.  So at some point in March(?) I go into the garage and I start corn and/or tall sunflower.  I have mesh networking working.  So my smartphone can send a message when I plant more plants than I think I need.  For the next little while, the server sends messages when it is time to water the corn/sunflower.  The time to plant outdoors depends on the outside soil temperature.  So, short of having radio temperature probes, the calendar needs to rely on the human for when the corn/sunflower get transplanted from hot box to garden.  I think you want to plant the (climbing) beans about the time when the sunflower is 6 inch tall.  I don't remember if you want to plant the squash at the same time.

You can guessstimate when to start harvesting the beans, but that depends on whether you are growing beans for the kitchen or for seed.  But really, you want heating degree days based on the base temperature of the beans in question.  And does this depend on precipitation/soil moisture?

And let's make this more complicated.  The corn you are growing is Stowels (sp?) evergreen corn.  So, if a frost is expected, you want to harvest the entire corn plant and let it finish in the garage.

I don't see how iCal could be extended to cover things like this.
 
Matt Coston
Posts: 136
11
  • Likes 1
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator

Sebastian Köln wrote:Why do you want to write new software?


As Gordon said, just for practice and curiosity.

On a complete tangent now - Gordon, I see you're interested in wireless sensors. Have you heard about a wireless technology called "LoRaWAN"? I only just learned about it. It is low power and low bandwidth but has a range of about 20 kilometers (under good conditions)

Here's a nice intro video for it.

 
master pollinator
Posts: 2815
Location: Toronto, Ontario
314
bee dog forest garden fungi homestead hugelkultur cooking rabbit trees urban wofati
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Ooh, that sounds useful!

I think that this sort of technology has great potential for intensively managing microbiomes within a larger permacultural system. You could plan so much more effectively if you had a grid network of sensors over a 20km area telling you specifics of anything from light levels over time to soil moisture levels, to, I don't know, Cation Exchange Capacity.

Then it's the sensor net sending regular updates to your planning software, with alerts going to your phone should the variables exit a comfortable range.

That data compiled over time would effectively be a record of observation, from a permacultural standpoint. You could go back in time, as it were, to examine aspects of the project you hadn't been working on at the time, but whose history might have just become relevant. Instead of sitting and observing a new space for another season, you could look at averages over the last twenty, let's say, and base your decision on extremely detailed local soil history.

That's definitely a part of the permacultural process with which I would accept automated assistance, allowing me to dig swales and hugelbeets and plant fruit hedges and perennial pasture alleys while a sensor net watches and records, and lets me know if something occurs that needs my attention.

-CK
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Off the top of my head, that sounds like Libelium and something called mote or waspmote.  Soemthing called Cooking Hacks seems to be a project site associated with them?  I may have spelled Libelium incorrect.

There is another similar technology called Grove.  I think that's from Seeed Studio (the word has too many e).

I haven't used either, but I can see where they might be useful.  They are sort of like already finished Arduino projects.

I was hoping to try the 900 MHz Xbee radio system.  To see if I could use it for getting corrections from a GPS base.  But, no time to build parts yet.

 
gardener
Posts: 562
Location: SoCal USA
96
bike cat composting toilet dog solar trees wofati
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
If a digital option is the goal, there are a lot of calendar and to-do list apps out there. Some can sync with other services and software, but for "online" I like ones where you can use a web site, as well as a mobile device and it will sync them to each other when you have internet access. That way if one device ever fails, you still have a backup.

Being able to schedule recurring events is rather handy, for example if you need to prune certain plants at different times or want to schedule harvest times. Then you can get reminders either the day of the event or some set time prior to give you some lead time.
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I looked some more at iCal and related.  I think that because iCal does not support lunar based calendars (such as used in Israel and Saudi Arabia), it is inappropriate for this.

The iCal article at Wikipedia, mentions a more complicated protocol which was dropped because it was too complicated.  It too appears to be inappropriate for this.

If I plant seed in the ground, I am not scheduling the harvest of said seed in the future.  Like a diary, I am committing to records, the time/day/conditions of the planting.  Programs monitoring the records, may at some point decide that they need to send a human to check if the plants need weeding, watering, pollination or other things.  It is possible to find sensors now, which can measure soil moisture.  In terms of scheduling a human to go water the plant, there is still the slight matter than mother nature may decide to water the plants herself (and where I live, that rain may in fact be hail or snow).  Maybe where you live, the government weather office has reasonable predictive ability.  But here, Environment Canada is probably only infinitesimally better than flipping a coin.  This probably because the eastern slopes of the Rocky Mountains are a bear to predict.

When I did my master's at C-MU, they computing science type had the pop machines connected to the LAN, so that they could query if there was enough orange crush in the pop machine.  And being a long time emacs person, I have run across coffee.el a couple of times (an implementation of the hypertext coffee protocol).  Both the coffee protocol and the tea protocol were published as April Fools joke RFCs, but I wonder if that might not be a basis for our scheduling needs?

Internet requests on the coffee protocol were apparently meant to start with:
coffee://

What would we want for URLs?
 ag://
cereal://
livestock://
aquaculture://
silvoculture://
other stuff?
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I am not trying to suggest permaculture submit an April Fools Day RFC.  The people that have written April Fools Day RFCs for IETF have typically gone to a lot of work to make something look reasonable/useful, and still be a joke.  But that should not preclude it from being a useful starting point.

How does one describe everything that can be done in agriculture?  My inclination is to suggest XML.  I have no doubt that other serialization methods such as JSON might not be more efficient.

Whether it is XML or other, has anyone ever developed a schema for what happens in agriculture?
 
pollinator
Posts: 2385
Location: Massachusetts, Zone:6/7, AHS:4, Rainfall:48in even Soil:SandyLoam pH6 Flat
125
forest garden solar
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
This would be a pretty big project.
Support
Nitrogen Fixers (Legume family, Alder, Seaberry Family, etc)
Insectory Pest Controllers (Mint family, Carrot Family, etc)
Soil Pest Controllers (Onion/Daffodils Family, etc)
Soil Aerators/Miners (Chicory, Borage Family, etc)
Biomass (Bio-char, straw, mulch, cover crop)
Innoculant (Worm Tea, Edible Mushroom, Beneficial Soil Life-fungi/bacteria/worms/nematodes)
Water Harvesters (city-size swales, regular swales, swale-keyline, swale-terrace, etc)

Plants
Edible/Main Plants and the amount of space they will claim at purchase, every year of growth and at maturity and reminder to thin every year.
N-Fixers at 90% coverage for establishment (mostly ground cover) down to 25% for "mature" food forest (canopy e.g honey locust) and reminders to thin
Logic to surround a plant/patch from a family (e.g. rose/apple) with plants from other families vs same family (other family=legume/mint/borage/cabbage/etc)
EcoRegion Preference and Min-Max Tolerance (Winter and Heat Zone, soil type, water requirements, etc)
Estimated Leafout/Bloom/Harvest time
Yield Amount and types (flowers, roots, fruits, seed, leaves, all, etc)

Mushroom
Oyster(straw) + Winecap(non-conifer woodchip) are pretty easy and just take care of themself once water and temperature variable match.
The above can be ground in soil or in grow towers/bag with automatic watering
There are other mushroom but they require more specific species of logs and or a
Green house like environment, with time to harvest/reflush, temp at different stage, humidity, preferred woodchip species


Animal
-Honey Bee:
They need a shed,
at least 3hives and a reminder when to check/harvest,
preference for Warre Bee Hive with a once a year Last Frost check/harvest, but other types and schedule exist

-Fish: Treated more like a greenhouse-setup
Automated Bulb+Electric Zapper for bug feed, automated BSF feed from compost, composting worms with calculation for stocking rate
mushroom compost from spent Mushroom Bag/Tower with calculation for stocking rate
Bio-Reactor/Sheltered Pond for Duckweed, Azolla with calculation for stocking rate
Store Bought Fish Feed with calculation for stocking rate and Kefir+Koji Fermentation
Fish preference/tolerance for Temp, Oxygen, Feed Rate, Nitrogen Levels etc
Monitors and Re-balancers for Temp-Heater, O2 Level-Aerator/Pump, Nitrates-Sand Bed/Plants, Sediment Filters
Pond Liner and pond usual at a depth of 3ft.
If the fish are raised in a natural pond at natural stocking rates, alot of the above can still be done, esp to funnel water into it, and filter incoming water

-Fowl
Alot of the same food source as the above, but they can be pastured or completely free range
Their bedding can incorporate bugs and compost in deep litter.
Bugs can be encourage in "pastures" esp woodchip "pastures"
Night time protection
Winter Egg Production (dusk/dawn lighting)
Coop system
Egg Collection Setup
Litter System

-Goat/Sheep/Cow
Rotational Grazing (Paddocks 30-45day rest)
Pasture Mix - Species Specific, EcoRegion Specific
Carrying Capacity
Fencing
Shelters
Milking
Killing
 
Posts: 167
Location: New Hampshire
15
forest garden hugelkultur tiny house
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
This all sounds to me like it needs some AI. The sort of stuff they're using for self-driving cars should be good enough. You'd record all the variables you think might be useful, and probably some you don't think might be but possibly could be, and it figures out how to use them to calculate what you need.

Recent AI advances are even getting to the point that the AIs will be able to tell us why.

So a combined permaculture Ai program could tell you when to plant your corn, tell you not to plant certain beans by the corn because of specific conditions where the corn was planted, when and which squash to plant by the corn, and when to harvest each of them. With proper sensors it could tell you to add water or certain types of compost.

With this sort of software you wouldn't have to remember all kinds of little details, and it could maximize production. Even idiots (or those who just don't care enough to learn) could follow the instructions. Imagine if your typical suburban family could grow a bunch of their own food using this, without having to learn all about it. Because lets be real, most people are never going to care enough about permaculture to learn all they'd need to in order to do it, but if they had a handy system that made it simple it would go a long way to getting folks on board.

Google just opened up their image AI API. My first thought was that a simple app using it to take pictures of maple leaves and/or bark could be used to figure out what kind of maple tree it is you're looking at. If it isn't a good sap producer it would make great firewood, so chop it. If it is, then record its position for tapping. You might even add a route planner to the app for collecting sap.
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Seeing AI mentioned did not raise my hopes.  I still remember about that language Lots of Inconsequential Silly Parentheses (and no, I have never written a eLISP program).

So, I went looking at what is going for AI on Linux.  And ignoring the stuff where you need TFLOPs or similar to do.

There is something going, which can run on as little as a Raspberry Pi 2, called Mycroft AI.

Hackster.io has a story titled "Hey Mycroft, Where is the International Space Station?", which talks about developing a "skill".  And that is sort of how I read all these problems that we find in permaculture.  Mycroft calls them skills.  So, if you get to be "expert" in one, you could sit down (if you have suffucient nerd skills ?) and write something, which lets some people solve a problem in permaculture.

Oh, a definition of "expert".  An "ex" is a has been, and a "spurt" is a drip under pressure.  

 
S Bengi
pollinator
Posts: 2385
Location: Massachusetts, Zone:6/7, AHS:4, Rainfall:48in even Soil:SandyLoam pH6 Flat
125
forest garden solar
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I dont think AI is needed (vision/Voice/permutation/etc skills)
I few data points will give it alot of information

User Inputted Address/GPS location, will give it your Heat Zone, Chill Hours, GDD rainfall amount and pattern eco-region, temp, soil type, size of your lot, ongoing temp data from internet, which will narrow down alot

At this point the software will do a base, support system layout and cost, based on selected area.

The user can then select what modules they want to add and their preference
Raised bed or regular vegetable gardenin
Fruit trees: Dwarf or Semi, usual or exotic
Nut Tree: 18ft or 40ft
Mushroom: Easy(Oyster/WineCap), Log or Greenhouse, yield wanted
Bee Hive: Easy Warre or Reg, yield wanted
Greenhouse: Size, Cold or Heated, Integrated or just vegetable, yield wanted slider
Egg/Chicken: Green House or Outside or Pastured, yield wanted slider
Fish:Heated Greenhouse or Outside, yield wanted
Milk Goat:
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
S Bengi

Are you talking about some specific piece of software which already exists?

 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
HackerNoon has a longish article on the predicate logic language behind Mycroft.

 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I am still looking at Mycroft issues, but ....

When Mycroft first came out, they also made the hardware thing Picroft, which was a Raspberry Pi computer to run the software.  I think it was a RPi-2.

In any event, Mycroft is hyping itself silly, because on Jan 25 they are introducing their Mark II hardware, which is another kickstarter thing I believe.  That said, there is lots of hype and bugger all as far as describing what the hardware is, or probably will be, or ....  But you can go follow the hype if you want.

While I can see speech to text (STT) as useful, I don't think it should be a core requirement.  Yes, if I am working with manure, I do not want to go typing on something.  Talking to the computer would be better then.  But in terms of scheduling stuff, I think most of the time the text can be input some other way than having a STT involved.

Where I am in looking at Mycroft, is the idea of running a server at your "home", and not using some corporate server somewhere.  Which suits me, if a mesh network can produce enough bandwidth.  Which to me, is a reason to keep STT out of it.



I'm not sure what this  message about Pie and permaculture all the time is about, but 3.141592653589793238462643383279502884197169399375105820974944
 
Gordon Haverland
Posts: 334
Location: Dawson Creek, BC, Canada
23
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
More rambling about mycroft.

In the past, if you needed speech to text for Mycroft, that was done by Google.  But Mycroft bundled things so that who originated the data was hidden (maybe, that's the claim).  Recently, they decided to switch to DeeppSpeech, which is (sort of) a Mozilla project.  At some level, it depends on a Google project called TensorFlow (just for software?).  In any event, if you want to run your own processor for doing speech to text, it seems that at the moment you have to use Nvidia GPUs, and a minimum of 8GB at that.  So, that is the GT 1070 and GT 1080 (if I go looking at Newegg).  In my (Canadian) dollars, the cheapest cards are $600.  Mycroft can still be the provider, regardless of what STT is used.  Or, you can set up your own STT server.  For me, the idea of a $600 GPU is expensive.

Some of the documentation was talking about tensor processors, and that they are starting to become available.  It could be a generalisation of an array processor, as tensors are a generalisation of arrays (sort of).  It may be they are talking some kind of FPGA.  I haven't looked into this.  FPGA can easily be $5000.  I remember looking a particular problem, and to solve with a GPU meant a GPU which uses more than 300W of electricity.  A FPGA to do the same work at basically the same speed, used 25W or so.  But that was a $5k (USD) FPGA.
 
You got style baby! More than this tiny ad:
permaculture bootcamp - learn permaculture through a little hard work
https://permies.com/wiki/bootcamp
  • Post Reply Bookmark Topic Watch Topic
  • New Topic
Boost this thread!