Ups! Almost forgot about the non-serious tag for that post. Here you have it:;) (+ extra help for individuals with extremely-poor-understanding skills: no, I am not afraid of being electro-shocked).
Additionally to my numerous posts with tons of grammar/spelling mistakes in both English and Spanish (my mother tongue), I do intentionally write non-existing words like "unmotivatedly". My question for you and the other supporters/executioners in the comments below: will I also be electro-shocked or should I expect a more painful punishment? BTW, nice full stop at the end of the sentence over there.
The differentiation STEM/humanities is kind of diffuse or, at least, not clearly defined across different countries and in different moments; in fact, the normal tendency seems to be moving away from black/white to greyer areas where both aspects might somehow complement each other. I am not sure about what your exact situation is, but mine is pretty hardcore in both: firstly (the order might also matter, BTW), fully focused on humanities and then on STEM (firstly engineering and then programming). When I was in each side I wasn't caring too much about the other one. I do perfectly understand that my position can be seen as a bit too extreme.
As per your post, it seems to me that you are more a humanities-kind of person. You might even work on STEM, but your attitude is humanities-prone. There are quite a few people working in programming from different backgrounds. There are also many positions closely-related to programming/engineering where different kinds of people work. In my original comment, I was referring to hardcore STEM/programming; to people really liking this entire world (not working on a somehow related setup and kind of liking all what it implies). Note that even people with advanced degrees related to all this might not be included there (not really liking/learning/understanding the given field and just passing through; in fact, this was pretty much me when being in humanities, that's why I don't consider myself a part of that, neither now nor back then when I studied it).
In summary, I don't think that you belong to that kind of (hardcore) STEM. I might also be wrong (and only your communication skills look like humanities material) and, in that case, you would be one of the first exceptions I know.
I firstly (, secondly, thirdly, etc.) watched it in Spanish and the translation lost in that specific bit (= "disidentes"). In general, it is very funny, but by using a pretty soft vocabulary. I guess that it was because of the times back then (1980), still very close to lots of prohibitions and censorship.
Although that statement certainly looks a bit too aggressive (+ prejudicious), the gap humanities/STEM is quite relevant (it might cut both ways if you wish). As a person who did pass through that gap in the past, I can confirm it. It is very difficult that the same person can really like both alternatives (I didn't like humanities and love engineering + programming): it isn't just knowledge; but also overall personality and expectations. Logically, everyone can learn to code or to write in whatever language; but really getting into it and becoming good enough is a different story.
It isn't just less prepared workers, but also clients, HR departments, managers, funding/financial entities and even dumbed-down programming environments/tests/solutions/expectations. I have no problem in recognising that, in the past, I let all this nonsense (because provokes everyone to lose) to somehow affect my work. It is also true that, unless you are lucky enough (or similar) to be in a quality-prone and supporting environment, things out there are quite tough: lots of offer at almost any price (+ quality/accountability) and lots of ignorance to make tons of horrible decisions. The high availability of information might also be provoking a wrong impression of control in people not precisely too aware about their actual (lack of) knowledge.
I don't think that coding camps or similarly simplistic/to-the-point/quick trainings are absolutely bad. Every single bit of knowledge can be positive when adequately applied. What is bad is misassessing what you can accomplish with that knowledge and what are the minimum requirements to get a proper product. The real problem are the final decision makers (HR/recruiters/managers for mis-hiring; clueless clients unaware about what they want/need; but also poor-knowledge programmers unaware about what they can/cannot do) who wrongly trust their unreliable knowledge (or other unreliable sources), who buy into the idea that everything is easy ("I read about a kid doing it, so has to be easy!") and contribute towards creating the current low-quality-, unreliability- and shortsighted-prone software development reality.
Further clarification to my previous post: the only serious part (= means exactly what is written) is that I don't use Node.js at all.
Help for those still not getting that I was being sarcastic: there are various clear indications about my true intention like liking a Monty Python video or including "Just in case: LOL.". In any case, the remaining text should ideally be more than enough to get my intention on account of the evidently nonsensical claims ("as everyone knows, fully agreeing with the code of conduct is a basic requirement before starting to use any programming language" or "become one of the most relevant episodes of our generation") and out-of-proportion intentions ("During the next months, I will be living in a cave..."). Hopefully, this will be my last post explaining what I consider my evident intention, although I will continue undoubtedly tagging all my not-completely-serious posts with additions like "LOL", meant to help those with serious understanding problems (to not waste my time).
As history taught us, even the tiniest difference should immediately provoke divisions; coming up with catchy names for each faction is also quite important (excellent work!). But this situation is harder than usual because, as everyone knows, fully agreeing with the code of conduct is a basic requirement before starting to use any programming language.
In fact and despite not using Node.js at all, I cannot ignore what is likely to become one of the most relevant episodes of our generation. During the next months, I will be living in a cave and hunting my food with my bare hands, just to get closer to mother nature such that I can reach the peace of mind required to even start thinking about what will be my position regarding a so extremely relevant issue.
Things are much more complicated than what you both think. A big proportion of the.NET Framework code has been public for a while now. As far as it is a huge reality (different languages, platforms, implementation types, versions, etc.), they divided it in different repositories. The one you are linking, CoreCLR, includes the most basic parts (e.g., contents of mscorlib.dll in Windows), but there are others like CoreFX (newer parts) or Roslyn (compilers and Visual Studio).
The.NET Framework has been systematically evolving and including more and more options and sub-types. Core refers to one of these classifications, although a quite big one: it aims to allow a somehow homogeneous management of the big number of supported platforms/formats. Here you can find a detailed description about it.
Thanks for the tip. I might test it next time (although relying on native.NET libraries is certainly an advantage when programming in.NET). In any case, it seems that my real problem might be almost-unsolvable on account of current speech-recognition approaches. The accent was problematic, but the biggest deal was the impossibility of recognising variable names (e.g., "var1", "function2", etc. all of them expected to be recognised as single words). Apparently, most of current speech-recognition approaches focus on existing words on the given language and try to match everything to them (in the best scenarios, "var1" recognised as "var" + "1"; and "function2" recognised as "function" and "2"). An alternative enabling the option of supporting pure phonetic (or spelling) would be excellent (I might even try to do something by my own). Anyway, I will see whenever I will decide to give a new shot at all this, what is very unlikely to happen before next year.
Yep, I understood that, and that's exactly what I was criticizing you for doing.
Yes, "some people" (hint: you) do have very poor understanding capabilities.
So, you are basically repeating what I said by inverting the targets (me rather than you). Synchronising words and reality isn't to important, right? In your mind, you have won this conversation, because this is all what it is about, isn't it? Going around saying random words (to anyone for any reason; the whole world is here waiting for you to come in for whatever reason and say whatever you feel like saying), with random meanings and randomly deciding what has been the output? Fascinating! LOL.
Sorry but I cannot believe that anyone (much less someone calling him/herself a programmer!!) isn't able to get a so simple idea! So here comes an additional explanation just in case: this wasn't about understanding a whole code as a conventional text, but about using speech-recognition (= me talking by using very simple and easy words) to trigger different actions (e.g., pointer moved to line X, Y text pasted in that part, bit Z removed, etc.). The only person thinking about creating a whole grammar to understand a programming language (?!) was you. I was only interested in having a reliable sub-system always understanding my instructions.
The ONLY PROBLEM which I found with this approach (other than that it was working perfectly) was that variable names aren't recognised because of not being valid (English) words (-> that's why I said that I wouldn't have even minded to spell them out; just the variable names which weren't recognised otherwise!!). If you want to help/criticise, please, focus on the actual problem rather than on your imagination: in-built speech recognition engine not being able to account for variable names (= invalid but similar to valid English words merged together).
[In another post, he mentions he was trying to actually dictate code....so who knows what he was actually doing.]
What about reading the numerous explanations which I wrote for you (because you needed much more than all the other commentators here to understand extremely simple ideas) and/or asking me? I have already wasted a ton of time with you and you don't seem to get even the basic ideas right, I wouldn't have minded to answer some TO-THE-POINT questions (already addressed in my other comments but which you weren't able to understand). In the future, please, try to avoid talking about me (actually lying as far as you transmitted an impression which has nothing to do with the reality as defined in the other comments) when I am not present.
Although you don't seem to be able to understand even simple concepts (I wrote a step by step example of what the application is doing and you are still not getting it?!), here it goes my last attempt at trying to go through that thick layer of poor understanding that seems to be around you, such that you can hopefully help 1 SINGLE IDEA RIGHT. Forget about grammar and all the rules which you apply to speech recognition in other contexts; forget about all what involves the syntax of a programming language; just assume English grammar by default and focus on answering this single question: how are you expecting me to create a set of grammar rules (or improvements on the default ones) allowing my application to recognise the extremely variable reality of variable names, as defined by examples like: "var1", "functionNew", "varString", "parseStringFromNumber"; that is: names with evident-for-people English meaning, but not expressly recognised in any dictionary. I want all of them to be recognised as single words, not as sentences ("var1" as "var" & "1" is wrong). Please, illustrate me and everyone else here: how can I use the in-built.NET speech-recognition engine to understand this kind of terms?
If you're still doing manual formatting/indentation in 2017, then you're doing it wrong, and you need to go back to school for a refresher.
(Hint: Programs like indent have been around for decades.)
If you take random words completely out of context and make up their meanings/my true intention, you are plainly having a discussion with yourself rather than with me. If you had read the remaining parts of that post you would have understood that I developed a small application to write code with my voice. For example, I say "insert void" (pause) "method1" and, in the corresponding file, it writes
void method1() { }
If I want to change the indentation (or the scope of the method is different and the default indentation too), I would use the corresponding command such that the application can perform the corresponding action (changing the indentation); FYI, that implementation is extremely straightforward (i.e., the difference between developing an application writing "{" or " {" is negligible). Do you see now the difference between using existing software to code (e.g., any IDE or editor since quite a few years ago) and developing a piece of software whose output is properly-formatted code; what my comment was about: the proper context which would have allowed you to have a conversation with me rather than with yourself?
Seriously, what is the problem of people with extremely-poor-understanding capabilities and their obsession of arbitrarily wasting my time with their nonsense? Aren't there better ways to spend time online than bothering me with their ridiculous expectations and concerns? Or why aren't they making a tiny effort to properly understand (even from their past huge amount of errors) such that they can have a slightly relevant contribution to someone else other than arbitrarily wasting time?
moving more and more towards cheating via rules like grammar because processing sound is too difficult
This seems like an spot-on summary of what I am starting to see that is the situation in this sub-field. I was expecting tools able to recognise random words by focusing on the sound aspect though.
And the problem is more difficult anyway. Try to dictate your program to another programmer: in my experience that never worked well.
If you have a proper underlying data-structure/-understanding in place there shouldn't be any problem on this front; at least, not for me as far as I developed this mostly for my personal use. I had still to think about good ways to pass reasonably long entities (e.g., a function with many arguments of different types), but I was quite happy with its performance during my preliminary tests. If random words (= variable names) were recognised properly, I would have certainly completed a first version for me to use in my work.
And to me "using simple words" sounded like a very strong arrogance red-flag, but whatever..
You misunderstood it. It was just an aggressive resource to minimise the chances of starting a long set of misunderstandings to nowhere. Although perhaps I did misjudge the situation and relied on it without being required.
use meaningful variable names so your code is more readable
I always write pretty meaningful English variable/method names in my code, but what is evident for a person is nonsense for a computer. For example, "variable1" (or worse, the more common "var1") is very difficult for a speech-recognition engine. And this was, as explained, precisely the main problem with that development and what made me stop it. Bear in mind that I was expecting this application to help me in my normal work, not to make it change the way in which I work (= redo all my variable-naming conventions).
I don't want to be rude, but I think that I have already spent long enough time explaining you what, honestly, I think that was pretty much implicit in my original post (at least, for someone with a reasonable good understanding about all this, precisely the only kind of audience who should be here commenting anyway). So, I hope that I have solved all your doubts and you don't need to ask me anything else.
I created that algorithm as a text parser/writer (+ communicating with the compiler) and relied on speech recognition only accessorily. Why spending time on letting the algorithm understand a context when it isn't required (+ would have increased the input complexity a lot)? This situation isn't equivalent to a conventional language full-text (with context, double-meaning, intention, etc.) understanding where the user has no participation, but to a mouse-pointer+keyboard emulator.
After starting, the application was kept in the background waiting for my commands. All the commands consisted in simple words and short sentences telling the application exactly what to do without it needing to understand previous/next order. For example: "open project" (small pause) "1", "open file" (small pause) "1", "scope" (small pause) "global", "go line" (small pause) "123", "insert method" (small pause) "method1" (small pause) "arg1" (small pause) "arg2", etc.
As explained to others, the problem I found cannot be solved via training: random variable names. Anything can be a variable name. I might not even have minded to spell it out, but that option wasn't available either (suggestion for those working on this front: why not including such an option?! It should be easy to implement and this kind of resources might be very helpful for weird-word scenarios or specific implementations like mine; or at least, a way to disable the language-dictionary searches and to focus on phonetic analysis).
What you are attempting to do is technologically possible
Sure it is possible, why would I have started to work on it otherwise? (I am an experienced programmer and, on top of everything, a sensible person with no interest in wasting my time for no reason). You (like others before too) are coming from a set of wrong assumptions regarding that specific development and my expectations: all what I wanted was a sub-system accurately recognising certain individual words, nothing else. I was the one meant to take care of all the context, different meanings, actions, etc. completely by my own. But the in-built speech recognition engine didn't fulfil that expectation.
I don't think you are using the one in the article
I worked on this around January/February and used the in-built speech-recognition engine of Windows/.NET. So, I was certainly not using what is being discussed in this article; and I haven't insinuated otherwise at any point.
Cool, bro. Can you hook me up with some access too?
No. LOL.
Because you must've used the internal MS Research software to do your anecdotal testing some months ago, since you've got an opinion on how good it is at doing its job.
As already explained to other poster with an equivalent (mis-)understanding, I was plainly sharing a relevant recent experience on this front to help people not too used to all this (e.g., myself 1 year ago) to get an idea about the current commercial reality (= way off 5%). You don't consider it relevant? Excellent! Ignore it. But, please, don't invent meanings or intentions which don't exist.
This means that you cannot throw meaningless words at it
This is precisely the reason why I stopped that development. It wasn't able to properly recognise a very important aspect of programming: random (variable) names.
I was sharing my personal experience on this front, not implying that the outputs of this research has anything to do with current commercial accuracy. I personally found kind of surprising the high number of errors (not too much into voice-based anything, but from what I see and read everywhere I was kind of expecting something different) and merely posted about that experience.
Doing what you want is a completely different thing and would use a completely different algorithm, so at the very least it as rather off-topic to this article
How is this true? I want my computer to write something from my speech and this is precisely what this is about. Imagine that I want to search for a specific brand with no meaning in English. This is also speech which is assumed to be recognised (at least, this was my approach). For example, Microsoft doesn't mean anything in English, but this is a quite commonly-used term.
You are being rather arrogant about it considering you very much didn't seem to understand the poster or why his criticism is valid.
I am not arrogant and none of my actions can reflect what I am not. I am plainly trying to be extremely clear (perhaps, aggressively clear) regarding my zero interest in dealing with people intending to have misunderstanding-, unmotivatedly-wasting-my-time and/or aggressive- (ironic, don't you think? LOL) prone attitudes. In any case and as explained above, I don't think that this criticism is valid. I clarified in my first comment that I do perfectly understand the complexity involved, but, as a final user of a speech-recognition piece of software , I don't care about that differentiation.
Slashdot Mirror
Ups! Almost forgot about the non-serious tag for that post. Here you have it: ;) (+ extra help for individuals with extremely-poor-understanding skills: no, I am not afraid of being electro-shocked).
Additionally to my numerous posts with tons of grammar/spelling mistakes in both English and Spanish (my mother tongue), I do intentionally write non-existing words like "unmotivatedly". My question for you and the other supporters/executioners in the comments below: will I also be electro-shocked or should I expect a more painful punishment? BTW, nice full stop at the end of the sentence over there.
The differentiation STEM/humanities is kind of diffuse or, at least, not clearly defined across different countries and in different moments; in fact, the normal tendency seems to be moving away from black/white to greyer areas where both aspects might somehow complement each other. I am not sure about what your exact situation is, but mine is pretty hardcore in both: firstly (the order might also matter, BTW), fully focused on humanities and then on STEM (firstly engineering and then programming). When I was in each side I wasn't caring too much about the other one. I do perfectly understand that my position can be seen as a bit too extreme.
As per your post, it seems to me that you are more a humanities-kind of person. You might even work on STEM, but your attitude is humanities-prone. There are quite a few people working in programming from different backgrounds. There are also many positions closely-related to programming/engineering where different kinds of people work. In my original comment, I was referring to hardcore STEM/programming; to people really liking this entire world (not working on a somehow related setup and kind of liking all what it implies). Note that even people with advanced degrees related to all this might not be included there (not really liking/learning/understanding the given field and just passing through; in fact, this was pretty much me when being in humanities, that's why I don't consider myself a part of that, neither now nor back then when I studied it).
In summary, I don't think that you belong to that kind of (hardcore) STEM. I might also be wrong (and only your communication skills look like humanities material) and, in that case, you would be one of the first exceptions I know.
Francis: Wankers.
I firstly (, secondly, thirdly, etc.) watched it in Spanish and the translation lost in that specific bit (= "disidentes"). In general, it is very funny, but by using a pretty soft vocabulary. I guess that it was because of the times back then (1980), still very close to lots of prohibitions and censorship.
Although that statement certainly looks a bit too aggressive (+ prejudicious), the gap humanities/STEM is quite relevant (it might cut both ways if you wish). As a person who did pass through that gap in the past, I can confirm it. It is very difficult that the same person can really like both alternatives (I didn't like humanities and love engineering + programming): it isn't just knowledge; but also overall personality and expectations. Logically, everyone can learn to code or to write in whatever language; but really getting into it and becoming good enough is a different story.
The title was meant to be: "Lots of money + demand [rightwards arrow] lots of stupidity".
It isn't just less prepared workers, but also clients, HR departments, managers, funding/financial entities and even dumbed-down programming environments/tests/solutions/expectations. I have no problem in recognising that, in the past, I let all this nonsense (because provokes everyone to lose) to somehow affect my work. It is also true that, unless you are lucky enough (or similar) to be in a quality-prone and supporting environment, things out there are quite tough: lots of offer at almost any price (+ quality/accountability) and lots of ignorance to make tons of horrible decisions. The high availability of information might also be provoking a wrong impression of control in people not precisely too aware about their actual (lack of) knowledge.
I don't think that coding camps or similarly simplistic/to-the-point/quick trainings are absolutely bad. Every single bit of knowledge can be positive when adequately applied. What is bad is misassessing what you can accomplish with that knowledge and what are the minimum requirements to get a proper product. The real problem are the final decision makers (HR/recruiters/managers for mis-hiring; clueless clients unaware about what they want/need; but also poor-knowledge programmers unaware about what they can/cannot do) who wrongly trust their unreliable knowledge (or other unreliable sources), who buy into the idea that everything is easy ("I read about a kid doing it, so has to be easy!") and contribute towards creating the current low-quality-, unreliability- and shortsighted-prone software development reality.
Further clarification to my previous post: the only serious part (= means exactly what is written) is that I don't use Node.js at all.
Help for those still not getting that I was being sarcastic: there are various clear indications about my true intention like liking a Monty Python video or including "Just in case: LOL.". In any case, the remaining text should ideally be more than enough to get my intention on account of the evidently nonsensical claims ("as everyone knows, fully agreeing with the code of conduct is a basic requirement before starting to use any programming language" or "become one of the most relevant episodes of our generation") and out-of-proportion intentions ("During the next months, I will be living in a cave..."). Hopefully, this will be my last post explaining what I consider my evident intention, although I will continue undoubtedly tagging all my not-completely-serious posts with additions like "LOL", meant to help those with serious understanding problems (to not waste my time).
As history taught us, even the tiniest difference should immediately provoke divisions; coming up with catchy names for each faction is also quite important (excellent work!). But this situation is harder than usual because, as everyone knows, fully agreeing with the code of conduct is a basic requirement before starting to use any programming language.
In fact and despite not using Node.js at all, I cannot ignore what is likely to become one of the most relevant episodes of our generation. During the next months, I will be living in a cave and hunting my food with my bare hands, just to get closer to mother nature such that I can reach the peace of mind required to even start thinking about what will be my position regarding a so extremely relevant issue.
Just in case: LOL.
Just posting to see what is wrong with my profile (it is still not showing my last comment?!).
Things are much more complicated than what you both think. A big proportion of the .NET Framework code has been public for a while now. As far as it is a huge reality (different languages, platforms, implementation types, versions, etc.), they divided it in different repositories. The one you are linking, CoreCLR, includes the most basic parts (e.g., contents of mscorlib.dll in Windows), but there are others like CoreFX (newer parts) or Roslyn (compilers and Visual Studio).
.NET Framework has been systematically evolving and including more and more options and sub-types. Core refers to one of these classifications, although a quite big one: it aims to allow a somehow homogeneous management of the big number of supported platforms/formats. Here you can find a detailed description about it.
The
Thanks for the tip. I might test it next time (although relying on native .NET libraries is certainly an advantage when programming in .NET). In any case, it seems that my real problem might be almost-unsolvable on account of current speech-recognition approaches. The accent was problematic, but the biggest deal was the impossibility of recognising variable names (e.g., "var1", "function2", etc. all of them expected to be recognised as single words). Apparently, most of current speech-recognition approaches focus on existing words on the given language and try to match everything to them (in the best scenarios, "var1" recognised as "var" + "1"; and "function2" recognised as "function" and "2"). An alternative enabling the option of supporting pure phonetic (or spelling) would be excellent (I might even try to do something by my own). Anyway, I will see whenever I will decide to give a new shot at all this, what is very unlikely to happen before next year.
Yep, I understood that, and that's exactly what I was criticizing you for doing. Yes, "some people" (hint: you) do have very poor understanding capabilities.
So, you are basically repeating what I said by inverting the targets (me rather than you). Synchronising words and reality isn't to important, right? In your mind, you have won this conversation, because this is all what it is about, isn't it? Going around saying random words (to anyone for any reason; the whole world is here waiting for you to come in for whatever reason and say whatever you feel like saying), with random meanings and randomly deciding what has been the output? Fascinating! LOL.
Sorry but I cannot believe that anyone (much less someone calling him/herself a programmer!!) isn't able to get a so simple idea! So here comes an additional explanation just in case: this wasn't about understanding a whole code as a conventional text, but about using speech-recognition (= me talking by using very simple and easy words) to trigger different actions (e.g., pointer moved to line X, Y text pasted in that part, bit Z removed, etc.). The only person thinking about creating a whole grammar to understand a programming language (?!) was you. I was only interested in having a reliable sub-system always understanding my instructions.
The ONLY PROBLEM which I found with this approach (other than that it was working perfectly) was that variable names aren't recognised because of not being valid (English) words (-> that's why I said that I wouldn't have even minded to spell them out; just the variable names which weren't recognised otherwise!!). If you want to help/criticise, please, focus on the actual problem rather than on your imagination: in-built speech recognition engine not being able to account for variable names (= invalid but similar to valid English words merged together).
[In another post, he mentions he was trying to actually dictate code....so who knows what he was actually doing.]
What about reading the numerous explanations which I wrote for you (because you needed much more than all the other commentators here to understand extremely simple ideas) and/or asking me? I have already wasted a ton of time with you and you don't seem to get even the basic ideas right, I wouldn't have minded to answer some TO-THE-POINT questions (already addressed in my other comments but which you weren't able to understand). In the future, please, try to avoid talking about me (actually lying as far as you transmitted an impression which has nothing to do with the reality as defined in the other comments) when I am not present.
.NET speech-recognition engine to understand this kind of terms?
Although you don't seem to be able to understand even simple concepts (I wrote a step by step example of what the application is doing and you are still not getting it?!), here it goes my last attempt at trying to go through that thick layer of poor understanding that seems to be around you, such that you can hopefully help 1 SINGLE IDEA RIGHT. Forget about grammar and all the rules which you apply to speech recognition in other contexts; forget about all what involves the syntax of a programming language; just assume English grammar by default and focus on answering this single question: how are you expecting me to create a set of grammar rules (or improvements on the default ones) allowing my application to recognise the extremely variable reality of variable names, as defined by examples like: "var1", "functionNew", "varString", "parseStringFromNumber"; that is: names with evident-for-people English meaning, but not expressly recognised in any dictionary. I want all of them to be recognised as single words, not as sentences ("var1" as "var" & "1" is wrong). Please, illustrate me and everyone else here: how can I use the in-built
If you're still doing manual formatting/indentation in 2017, then you're doing it wrong, and you need to go back to school for a refresher. (Hint: Programs like indent have been around for decades.)
If you take random words completely out of context and make up their meanings/my true intention, you are plainly having a discussion with yourself rather than with me. If you had read the remaining parts of that post you would have understood that I developed a small application to write code with my voice. For example, I say "insert void" (pause) "method1" and, in the corresponding file, it writes
void method1()
{
}
If I want to change the indentation (or the scope of the method is different and the default indentation too), I would use the corresponding command such that the application can perform the corresponding action (changing the indentation); FYI, that implementation is extremely straightforward (i.e., the difference between developing an application writing "{" or " {" is negligible). Do you see now the difference between using existing software to code (e.g., any IDE or editor since quite a few years ago) and developing a piece of software whose output is properly-formatted code; what my comment was about: the proper context which would have allowed you to have a conversation with me rather than with yourself?
Seriously, what is the problem of people with extremely-poor-understanding capabilities and their obsession of arbitrarily wasting my time with their nonsense? Aren't there better ways to spend time online than bothering me with their ridiculous expectations and concerns? Or why aren't they making a tiny effort to properly understand (even from their past huge amount of errors) such that they can have a slightly relevant contribution to someone else other than arbitrarily wasting time?
moving more and more towards cheating via rules like grammar because processing sound is too difficult
This seems like an spot-on summary of what I am starting to see that is the situation in this sub-field. I was expecting tools able to recognise random words by focusing on the sound aspect though.
And the problem is more difficult anyway. Try to dictate your program to another programmer: in my experience that never worked well.
If you have a proper underlying data-structure/-understanding in place there shouldn't be any problem on this front; at least, not for me as far as I developed this mostly for my personal use. I had still to think about good ways to pass reasonably long entities (e.g., a function with many arguments of different types), but I was quite happy with its performance during my preliminary tests. If random words (= variable names) were recognised properly, I would have certainly completed a first version for me to use in my work.
And to me "using simple words" sounded like a very strong arrogance red-flag, but whatever..
You misunderstood it. It was just an aggressive resource to minimise the chances of starting a long set of misunderstandings to nowhere. Although perhaps I did misjudge the situation and relied on it without being required.
use meaningful variable names so your code is more readable
I always write pretty meaningful English variable/method names in my code, but what is evident for a person is nonsense for a computer. For example, "variable1" (or worse, the more common "var1") is very difficult for a speech-recognition engine. And this was, as explained, precisely the main problem with that development and what made me stop it. Bear in mind that I was expecting this application to help me in my normal work, not to make it change the way in which I work (= redo all my variable-naming conventions).
I don't want to be rude, but I think that I have already spent long enough time explaining you what, honestly, I think that was pretty much implicit in my original post (at least, for someone with a reasonable good understanding about all this, precisely the only kind of audience who should be here commenting anyway). So, I hope that I have solved all your doubts and you don't need to ask me anything else.
Did you build your own grammar?
I created that algorithm as a text parser/writer (+ communicating with the compiler) and relied on speech recognition only accessorily. Why spending time on letting the algorithm understand a context when it isn't required (+ would have increased the input complexity a lot)? This situation isn't equivalent to a conventional language full-text (with context, double-meaning, intention, etc.) understanding where the user has no participation, but to a mouse-pointer+keyboard emulator.
After starting, the application was kept in the background waiting for my commands. All the commands consisted in simple words and short sentences telling the application exactly what to do without it needing to understand previous/next order. For example: "open project" (small pause) "1", "open file" (small pause) "1", "scope" (small pause) "global", "go line" (small pause) "123", "insert method" (small pause) "method1" (small pause) "arg1" (small pause) "arg2", etc.
What you are attempting to do is technologically possible
Sure it is possible, why would I have started to work on it otherwise? (I am an experienced programmer and, on top of everything, a sensible person with no interest in wasting my time for no reason). You (like others before too) are coming from a set of wrong assumptions regarding that specific development and my expectations: all what I wanted was a sub-system accurately recognising certain individual words, nothing else. I was the one meant to take care of all the context, different meanings, actions, etc. completely by my own. But the in-built speech recognition engine didn't fulfil that expectation.
I don't think you are using the one in the article
I worked on this around January/February and used the in-built speech-recognition engine of Windows/.NET. So, I was certainly not using what is being discussed in this article; and I haven't insinuated otherwise at any point.
Cool, bro. Can you hook me up with some access too?
No. LOL.
Because you must've used the internal MS Research software to do your anecdotal testing some months ago, since you've got an opinion on how good it is at doing its job.
As already explained to other poster with an equivalent (mis-)understanding, I was plainly sharing a relevant recent experience on this front to help people not too used to all this (e.g., myself 1 year ago) to get an idea about the current commercial reality (= way off 5%). You don't consider it relevant? Excellent! Ignore it. But, please, don't invent meanings or intentions which don't exist.
Hijo mio, soy Venezolano.
Vale, papi, lo de "william" me despistó (pero supongo que yo tampoco puedo hablar mucho en ese frente). LOL.
Si eres español, mis condolencias por lo de Barcelona.
Gracias. Cómo repagar ese gesto a un venezolano? Sé que algo había pero no caigo yo ahora... LOL
En español de españa: a mi me la suda...
Ahí estamos! Hablando con claridad!
Encantado y a más ver.
This means that you cannot throw meaningless words at it
This is precisely the reason why I stopped that development. It wasn't able to properly recognise a very important aspect of programming: random (variable) names.
I was sharing my personal experience on this front, not implying that the outputs of this research has anything to do with current commercial accuracy. I personally found kind of surprising the high number of errors (not too much into voice-based anything, but from what I see and read everywhere I was kind of expecting something different) and merely posted about that experience.
Doing what you want is a completely different thing and would use a completely different algorithm, so at the very least it as rather off-topic to this article
How is this true? I want my computer to write something from my speech and this is precisely what this is about. Imagine that I want to search for a specific brand with no meaning in English. This is also speech which is assumed to be recognised (at least, this was my approach). For example, Microsoft doesn't mean anything in English, but this is a quite commonly-used term.
You are being rather arrogant about it considering you very much didn't seem to understand the poster or why his criticism is valid.
I am not arrogant and none of my actions can reflect what I am not. I am plainly trying to be extremely clear (perhaps, aggressively clear) regarding my zero interest in dealing with people intending to have misunderstanding-, unmotivatedly-wasting-my-time and/or aggressive- (ironic, don't you think? LOL) prone attitudes. In any case and as explained above, I don't think that this criticism is valid. I clarified in my first comment that I do perfectly understand the complexity involved, but, as a final user of a speech-recognition piece of software , I don't care about that differentiation.