Talk:Using a speech engine to highlight words

Problems

I think their may be problems with the task as their will be too few languages with in-built speech-recognition features. I suggest we delete the task and the original task creator add the suggestion to Rosetta Code:Village Pump/Suggest a programming task maybe?

Unless two or more languages have the ability, the task will devolve into ways to call an external library - for which we already have tasks, and it will hardly help comparing languages.

In summary, I think the task may break the guidelines section "Don't require exceedingly rare features.". --Paddy3118 12:25, 4 April 2011 (UTC)

I don't have a problem with tasks which wind up primarily calling out to libraries. Seeing how a language interacts with a particular library isn't necessarily useful in comparing a language to another language, but it is still useful for users of that language and users of that library, and may be useful in comparing libraries. Utility to users of a particular language is a significant language-community motivator for adding code to the wiki, and I've seen several languages whose communities get together to add to RC for that purpose. As for comparisons between libraries...that's something I've wanted RC to be able to cover for years. Like languages, libraries are another tool where active competition and code and idea development occur.

The bit about "don't require exceedingly rare features" was not intended to excluding calling outside the language's native semantic space. It was more intended to deal with tasks like Proof which have a very narrow definition and are written to be intolerant of non-native features. --Michael Mol 12:59, 4 April 2011 (UTC)

Thanks for the clarification Michael. --Paddy3118 14:01, 4 April 2011 (UTC)

Yeah, sorry the article is about computerized speech, not speech recognition. The task was badly worked, because I rushed this article through on a dinner time, whilst demostrating Rosetta Code to a colleague at work. I have fixed this now.

I am not entirely happy with the title. Maybe this should read "Highlight each word as is it spoken by the speech engine".

Markhobley 16:47, 4 April 2011 (UTC)

I suspect that we ought to do a more basic speech-related task first, such as saying a constant string like “Hello, world!” Without being able to do that, there's no way to take on a much more challenging task like this one. –Donal Fellows 17:21, 24 April 2011 (UTC)

Remove and add to suggested tasks list?

It has been several weeks without the first language implementation. Without one, what is the difference between this task and one on the long list of requests for tasks?

Maybe it should be retired there? --Paddy3118 18:17, 24 April 2011 (UTC)

It is easy enough to write an implementation. Apparently VB6 has this facility built into the speech engine, enabling it to be achieved by a programmer with an beginner/intermediate level of skill. (I am not a developer for that platform though, so I haven't got any sample code for this at this time.)

For other engines, we can emulate this with a parse, highlight and speak loop. As long as we can control the cursor, this should be easy. I'll knock some code together for other languages as time permits. Markhobley 18:48, 24 April 2011 (UTC)