Major architectural change. Before updating you will need to run
pip install dfly-breatheI have entirely replaced the old caster-based CCR backend with Breathe, a library I've been working on. Should make it much easier to add support for new apps etc, and I have fully fleshed out the NestedRule concept which was originally prototyped here. Please let me know if there are any issues.
Well basically I wanted rules that were like:
Integral from <stuff1> to <stuff2>Where stuff was a sequence of any other ccr commands. Nested rules were the first implementation based on Caster, breathe allows this by allowing the user to add custom top-level rules. The mathfly examples are in e.g. lyx.py at the bottom but may be slightly difficult to understand at a glance, there are clearer examples on the breathe readme which show greedy dictation and optional repeat values using the same feature.
CommandsRef extra which references a sequence of commands and then include an Exec action to execute the sequence
say <text> command, which should allow you to input text without using Dragon dictation
@mrob95 Not urgent as it only happens a few times an hour, but I wanted to report an error I'm getting with some of the longer strings of math dictation. In particular, I have latex-maths enabled and I'm getting some warnings in the Python output window. The relevant utterances also are simply not entered into my document. Here are a few examples:
WARNING:engine:Grammar MergeRule(Merged42Ma): failed to decode recognition (u'equals', u'uniform', u'sub', u'one', u'right', u'equals', u'uniform', u'sub', u'two', u'right', u'equals', u'uniform', u'sub', u'three', u'right', u'ampersand', u'equals').and
WARNING:engine:Grammar MergeRule(Merged15Ma): failed to decode recognition (u'alpha', u'plus', u'x-ray', u'minus', u'one', u'right', u'prekris', u'one', u'minus', u'papa', u'right', u'super', u'script', u'eight', u'up', u'plus', u'november', u'minus', u'x-ray', u'minus', u'one', u'right').Note that in the first example all of the component words are recognized correctly, while in the second example the word 'bravo' was misrecognized as 'eight up'. I believe even the misrecognition should have yielded a valid sequence of commands, however (if a somewhat silly one).
Like I said, not urgent, but probably a bug worth looking at some point. Let me know if I can provide any more information. Thanks.
Hi Anna, firstly I'm impressed that you can hold such a long string of commands in your head at once ^^. I have always needed a pause for thought after a few commands.
I am not 100% sure that this is the issue, but one thing you could try is bumping up the value of max_ccr_repetitions in settings.toml a bit. I don't think there is any science behind the current maximum of 16. Someone chose it a long time ago as a number that was big enough to rarely cause issues while also not triggering grammar complexity errors, and it stuck.
I'm talking about working through a problem where you are not sure exactly what strategy to use in might need to try multiple things before finding the correct approach. (So, not a situation in a class where you basically know what to do and just have to mechanically follow the steps to do it.)
I have used math fly exclusively for dictating LaTeX. This is great for presenting my work, but I have never been able to use LaTeX when actually solving math problems. I just can't keep track of enough information in my head and/or is too hard to find the information I need in my source document and/or I am distracted by compilation errors.
Of course, LaTeX is (to oversimplify) basically a markup language. LyX and Scientific Notebook are significantly different in that you could at least see the math that you would dictating in real time.
As pain in my hands/wrist/arms/shoulders can make handwriting difficult, I'm considering whether it is worth the time investment (and, for Scientific Notebook, financial investment) to become proficient with mathfly + LyX or Scientific Notebook. I would be interested to hear anyone's experiences.
I definitely think it's possible - I can't write at all and used scientific notebook throughout my degree for all homeworks, exams, etc. I never needed extra time in exams so it's not a huge amount slower than handwriting I don't think. There is still a bit of mental overhead involved in mapping things to commands in your head and fixing errors but it's manageable when the output is rendered instantly.
The LyX grammar is fairly similar to the LateX one so it probably wouldn't be too much of an investment to try it out and see how it fits. Personally I prefer scientific notebook 5.5 but I think the only way to get hold of it is to buy mathtalk - version 6 is much worse for voice input imo. @esc123 has used both I think so they may be able to give an opinion on whether it would be worthwhile.
I also have trouble with working through a problem in latex. But I think with lots of grammar customization it should work. Just the difference going from "x-ray underscore number one" to "x-ray score one" already helped a lot.
I think lyx is much better suited as you can at least see what you are doing. For me the biggest hurdle is actually navigating within a formula in lyx. And I have not found a work around for that yet...
Would definitely be worth a try I think:
https://www.mackichan.com/index.html?products/dnloadreq55.html~mainFrame
There is none of the "nested boxes" nonsense you get in lyx
@mrob95 I would definitely be interested to hear if you find a good free alternative to scientific notebook.
@knork_gitlab I don't think grammar customization can entirely solve the problem of it being hard to work through problems in LaTeX. I am so "fluent" in the mathfly LaTeX grammar (with a few personal modifications) that I can transcribe entire mathematical expression/equations from a written form with essentially no pauses. (In fact, I may have broken the grammar complexity rules by dictating too much without pausing. See discussion above.) So, I have absolutely no challenges taking an equation in my head and inputting it to the software. That mapping is so ingrained that I sometimes accidentally use a phonetic alphabet when speaking about math to others. So, I'm really not convinced that needing to remember how to speak expressions is a limiting factor for me. (Granted, I am pretty sure I have literally spent hundreds of hours dictating math with math fly...)
Where I run into trouble is the need to then operate on those expressions to create new expressions. When my expressions get long, it can be really hard to find the parts that I need with in potentially multiple lines of text in my text editor. This is especially true when there are many nested functions and it's hard to keep track of which curly braces go together. I think my issue really is related to the fact that written mathematical notation serves a purpose as essentially spare memory. It is really hard to hold a long expression in working memory, so the notation helps us keep track of all of the parts of an expression/equation. When my expressions get complex enough/long enough that I can't easily read and interpret the LaTeX source, I essentially lose the function of that extra working memory.
This does pose the question of whether an editor that does real-time compilation could be helpful. I suspect it could be, but then we have to worry about how that editor will behave with Dragon. Of course, the math fly commands themselves are just keystrokes and should be fine, but actual mathematical proofs also contain sentences. Having to open the dictation box (or remembering the "say" command) every time I need to place an 'and' between equations sounds pretty terrible. But it might be an idea worth looking into nonetheless.