Ultimate typing tester?

Thought it might be interesting to look at what would be nice to have on typing test pages online as I’ve never found any one that did all I like. Not particularly emphasizing training where you may only do home row characters and work up. This is more for general testing with some possible limitations on what you have to type.

So, just writing these things as I think of them, it should allow for various options:

  1. Selection of what is included. Basic would be just words. No punctuation, symbols, or numbers. You might even leave out capital letters. But then check boxes would let you include any of those things you want. For me, I’d do basic punctuation and capitals. Occasionally numbers. For symbols, I’d probably like to be able to fine tune it. I mean, sure, dollar sign, percent, ampersand, equals, etc. But, no, I probably wouldn’t care about square or curly brackets or a number of other less common things for my use. But if I was great at all the rest, sure, later I might add those.

  2. I like a large, running display of my wpm, how much time is left, and my accuracy percentage. Maybe some would not. So give us the choice of what to show - but with a big enough display for what does show that I can see it easily while still focusing on the words. Speedtypingonline has a nice display of this information.

  3. Time internals ranging from 30 seconds to at least 5 minutes - even better, just let you enter the exact time you want. If you want 27 seconds, why not?

  4. Let me have an option to see the letters I type appear right on the words they show, clearly highlighting errors (in a color that doesn’t make it hard to tell if you are color blind!) OR type in a box under their text box. I prefer the former usually.

  5. Have an option to either keep typing if you make an error or have it force you to correct any error first. I prefer the former myself since I think it is more realistic. It must also allow you to edit back multiple words. Some of these, once you hit a space, you can’t go back and edit it! These things should ALWAYS at least let you chose to work in a real world situation.

  6. If the text is random, let it have an option to react to errors intelligently. For example, today I was taking a test and missed an “L” (but lower case). By coincidence I noticed I got a fair number of words with that letter and in a similar position (typically the second letter of the word). So, it would be cool if this was done on purpose in random text. It would note the letter you missed and, at least, the character that preceded it as well as the full word. Then give you the same or similar two letter combinations to get you to focus on it. When I started seeing myself missing that letter a bunch of times, I knew my hand position had gotten out of whack.

  7. Results should list what words you missed and how you typed them. On the: http://typing-speed-test.aoeu.eu/?lang=en
    website, they do this and it is great for catching patterns of errors. A couple weeks ago I was seeing the vast majority of my errors were hitting C instead of T and T instead of W. I still miss those the most, but they don’t constitute the majority of my errors anymore.

  8. The speedtypingonline website also lists the 3 characters you miss the most, are most inaccurate on, type fastest and type slowest. I think these could be refined a bit, but it is useful information, especially the actual ones you miss.

  9. Final results should include total characters typed, number of missed characters, how many words were missed (since you could make two individual errors in one word so I want both calculations), and raw speed. Oh, and besides how many errors you made in total, but how many were corrected or not.

  10. Personally I don’t think a corrected error should subtract from your wpm score. You lose speed because of the time it takes for the correction, but hey, it’s fixed. Why should it count against you twice - just include it in the error number, but only unfixed errors would affect your wpm score.

  11. Some sites let you specify what letters you want to focus on. Trouble is, when I enter letters, I get ONLY those letters - not actual words! This isn’t good. We learn to type words. If you are new to touch typing, it is the spelling of the words that slows you down a lot. When you get good, the your fingers just automatically type out words, not letters. So, let people specify letters, but then provide words that CONTAIN those letters - and not every word needs to. Just so you get far more of those letters than normal. At least give an option to do it this way instead of only getting the random letters.

  12. Of course, let you enter your own text to use for text - but let you keep as many as you like which can either be done over and over or are added to the random selections of quotes the app would also provide.

  13. Just remembered another - need to adjust for errors then following letters are correct. This isn’t a problem if you type one error so the number of characters doesn’t change. But what if you have a three letter word but type 2 or 4? On some tests sites, everything you type after this will be wrong even if the order of letters is correct. At the very least they should do what some do - reset any time you hit a space. But even this could be better. For example, occasionally I may not hit space tell so two words are connected - so every letter after that missed space will be marked wrong. Also, when you do hit space after the end of the second word, the program now thinks you are ready to START the preceding word while you are actually on the next one. So, for one space error, you could conceivably have everything marked as wrong after it! The computer should be able to handle this better.

As soon as WT finishes the hardware and firmware (for Android too), they can get to work on this project! Besides a website, it would be really nice to have it as a computer, phone, tablet program since the web can sometimes get slight delays in displaying what you type. I find even minuscule delays throws me off when I’m watching the screen.

1 Like

Have you tried keybr.com? Gets some of your suggestion right (some totally not). It’s what I used to learn colemak recently: the developer has just rewritten everything from scratch to improve the codebase so it’s the best moment to ask for features!

1 Like

With all of that it’s not a tester anymore, it’ll be a coach helping you to train for the typing Olympics :smile:

Yeah, I tried keybr.com. Hate the mark they use for spaces! There is some other site that makes the space too narrow and when giving random letters (which I don’t like anyway, sometimes its hard to tell at a glance is something is a space!

Also, the site doesn’t seem to let me move past a mistake even though I tried to set it to move forward after an error. The cursor will then continue to move - but so does the text!

I love keybr and have been planning on using it from the beginning once my TB arrives. Great tool.

I’ve been using it for awhile today and it does have some nice stuff, but some annoyances.

I already mentioned the weird character they use for spaces, which just makes things look cluttered to me.

Unless you do your own text, you don’t get many actual words but rather syllables that are somewhat readable like words. But you mostly end up with partial words that way when the mind wants to automatically type the whole word. I think that is fine as an option, but would much prefer full words.

While you can enter your own text, unless I’m missing something, you can only have one thing at a time to work on. I’d like to gradually keep adding things which the app would then choose from at random (but with an option to specify which one(s) you want to temporarily limit it to.

I just tried using my own text and have two issues as you reach the end of the line of text showing - first, you can’t see the next word until you type the last character on screen! That isn’t natural! Neither is the fact that when you have the next line of text appear, you don’t have to have ever typed the space between it and the word that preceded it on the prior line!

I don’t see any timer option either - you just keep typing the same thing over and over without a time limit or even a the time whenever you just decide to stop.

Let’s see, also the speed can only be CPM or WPM - not both. I like both. And it only updates when you finish each line. It also seems that each line gets it’s own score rather than getting the average as you type. I see there is some data under Profile, but I shouldn’t have to go their to get such basic info - and it doesn’t seem to give scores for individual tests.

I can not get the thing to pull data in from a website. I’ve tried with safari as well as an Opera Developer browser.

Another thing, if you choose to allow punctuation and you aren’t using your own text, it seems you get plenty of the less common ones - don’t like that at all! Needs a list you can select from.

OTOH, it does provide a ton of data in your profile - which I haven’t had time to study yet to see how useful it is.

I agree, the way it handles mistakes is kind of frustrating, especially at higher speeds. It sometimes sticks on the letter I typed wrong. Other times I find myself 5 letters past the error and have to stop and figure out where the cursor is so I can pick up in the correct spot (vs continuing to make errors because of where I think the cursor is).

It had my typing speed varying from 74 to 130, depending on the sentence I was typing and whether a mistake stopped me long enough to catch my place.

Dang, I don’t know how people get that fast!

Actually you can get all the data for each line (a session I think he calls it) to get the whole timing thing to analyse on your own, which is damn cool. Oh, and I got to 111 wpm peak in colemak, which is now officially faster than my qwerty ever was (I never had peaks of more than 90, and my sustained speed was 70, which is the same as now, but now feels way more relaxed)

I may not be understanding how their results are done so feel free to elaborate! I assume each line is treated as a test? Which makes for very short tests! I just haven’t used it enough to figure it all out. I generally consider a one minute test to be the minimal time for a meaningful test. Longer is better, but I get too frustrated to type longer before knowing how I did! So I’d rather to 5 one minute tests and note the trend rather than one 5 minute test.

Yup, each line is a test, and you actually have the timing (I think!) for each letter, for each word in each test. So you can find out how fast you were typing the last t in your post, specifically.

It’s a little bit weird, as far as I see it (I agree with you on meaningfulness of data through longer periods).

And I’d like to type on my textblade already, btw, the mac keyboard is slightly not wireless :frowning:

Pardon my ignorance, but how useful will keybr.com be if it performs its exercises against a standard QWERTY keyboard and not the TB layout?

You can set it to dvorak. There may be some quirks in the display since TB has fewer rows, but I don’t think that is a major issue. At least not worse than any other I’ve seen.

Maybe some folks can explain some of the stuff from keybr.com because a bunch of it makes no sense to me.

Here is part of what I got after my last run through:

Average typing speed:352cpm
Best typing speed:343cpm
Confidence level:0.98
Learning rate:~0.025cpm/sample, ~5.613cpm/hour

Let’s start with the obvious - how can my average typing speed be faster than my best??? I’ve also noticed that these two numbers are often the same. I also have doubts about the accuracy. After all, these numbers would represent about 69-70 wpm. Yeah, I can type that fast - but most of my tests lately have had a bunch of “q” characters which really slow me down, especially since they mostly aren’t real words so I have to think more about letters rather than the more efficient “words”. So I don’t see how my speed could be this fast.

Is “confidence level” like percent accuracy or based on how long you delay between letters or what?

I don’t get this learning rate thing either. I’d normally guess it shows how much you are improving (hopefully) over time.

The first one says about 0.025 cpm per, I guess, test. Is this based on your scores from the start compared to the average of X number of more recent tests? I can think of a lot of ways this could be done. I tried typing one real slow so I’m assuming it isn’t based just on the last test since it would probably have been slower than normal, but it doesn’t show it as negative.

Similar questions about the improvement per hour.

And this is just the most basic stuff. There are lots of other things I don’t get.

1 Like

This seems to be a bug that has not been fixed, I assume it’s just reversed

I always suspected it’s the confidence level in the learning rate you have, but not sure.

It seems to be based on data from the beginning of time, but again I’m not a 100% sure about it.

It’s really cryptic at many things indeed.

1 Like

Also, the numbers you’re quoting are for a specific letter, don’t they?

Looks like something is broken in reporting since… Two days ago, last time I checked:

23 is not 128, this is data for the letter E:

But quite often the numbers would be exactly the same - average and fastest shouldn’t be that way unless there was only one test! And that is particularly fast for a “q”.

Just noticed that, after reading your post. Still seems like odd results. I’ve had a lot of “q” for awhile now. Checked all the letters across the top by hovering over them. All have a confidence level of “1” except the q which is presently 0.98. I guess it is going to keep loading me up on that until it also reaches “1”.

Haven’t figure out the “delta” references for letters either.