Future ideas

[kdwhite]

Brien kind of kicked this off in another post. at some point we'll begin work on new version of TextAloud, so I'd love to hear any and all ideas. Nothing is too big or small, nothing too stupid. So if you have any ideas, let's hear them.

[BrienMalone]

Thanks for the responses!

I have to admit, I didn't fully consider your target market (regarding interface design and page handling ideas) I've started some quick photoshop work on an interface suggestion - hope to have more time this weekend.

Skins
Hey - How about a skinable interface? That'd be pretty spectacular... (of course, low on the priority list. Only a handful of us nerds would do anything with it.)

Tunes
And this is really in the "nice-to-have" pile... I listen to nice stale books via TTS in my car on the way to work (topics like Business Analysis, Software Project Management, Fuzzy Expert Systems,...etc.) To punch them up a little, I'll play music in the background using another media player.

How about a media player component that will allow you to associate a track or group of tracks with articles... If the article runs out before the music does, it can fade out, then fade into the next page.

(I think I'm only half joking with this)

Survey
You know... PHPbb allows you to conduct polls. Perhaps once you reach the cut-off point for 3.0 ideas and have narrowed the list down to what is possible, plausible and practical, you can put the feature candidates up on the forum for a vote... ranking them on a 1-10 scale according to how interested the TA user community would be in seeing that feature included.

Just a thought!

Cheers!

[kdwhite]

Skins

Try View->Skins

Tunes

This is actually on the list, the way to specify an audio file that would kind of get mixed in with the text audio. Not sure how high a priority it will be, but we've heard it a few times.

Survey

I agree, this is something we'll try to do.

[SFCurley]

I'm sure this is on the list already because it's been previously discussed in the forums, but here it is: the ability to set the order of regular expressions in pronounciation editor.

[Bunger Henry]

I would have to say the number one thing I'm looking for is an Advanced Pronunciation Editor that works for all voices, i.e. Heather, Ryan, et al.

With Heather, sometimes I really struggle to find a way to make it pronounce a word correctly with the Basic Pronunciation Editor.

Try to make it say the word "babbies," where the "a" is flat as in "babble." Very hard. Eventually you will figure out that "bab ease" is the closest you can get, even though it is not quite as smooth as a single word.

Now try to make it say the word "Stri-Dex," where the first half of the word rhymes with "try." I still haven't figured that one out!

Please add Advanced Pronunciator to your list.

[D.Leikin]

At the moment, these problems are easily resolved by introducing a proper pronunciation string directly into Acapela’s native vocabulary file

…\Program files\NextUp-Acapela\data\exc_usa.txt

For example, you can insert the line

Stri-dex : [s"t"r"ai"d"e"k"s"] (NNP) /i

into the body of this file to get this word pronounced correctly.

Two tips to make it work:

(i) uncheck the read-only box in the file’s properties,
(ii) restart the speech cube after you have completed editing and saved the file.

Cheers!

[D.Leikin]

…have forgotten to formulate an idea. I am nearly sure you have it on the list, still…

Different voices have different lexicon files using different phoneme tables, e.g. those in NeoSpeech’s userdict_eng.csv, Acapela’s exc_usa.txt, etc.

Wouldn’t it be smart if these files could be handled directly by the Advanced pronunciation editor using some unified (and, if possible, intuitively understandable) phoneme table?

This will not only make it possible to handle pronunciation problems which, presently can not be resolved in TA (like that in the previous post), but also make it easy to correct pronunciation for a single voice leaving all the rest intact. This seems quite important as different voices have different pronunciation problems.

[Bunger Henry]

At the moment, these problems are easily resolved by introducing a proper pronunciation string directly into Acapela’s native vocabulary file

I looked at that file and saw a lot of unfamiliar phoneme codes. Do you know where I can find a complete list of those codes and their associated sounds?

[D.Leikin]

Unfortunately, I don’t know where to get them, sorry about that and about my off-topic post here too. However, I guess things are not all that difficult if you just peer at the examples within the file itself. For example, (") – separates the phonemes, (*) – the accent, (/i) – ignore case, etc. The phonemes are pretty lucid, many of them being explained in the lines of the file. BTW, did [s"t"r"ai"d"e"k"s"] work for you?

[Bunger Henry]

I haven't tried experimenting with that file yet. I wonder if kdwhite would be willing to ask his contact at Acapela for a table of the phonetic codes. I would be willing to experiment with the file some if I had a list of the codes.

[kdwhite]

We'll get to it at some point. At the moment the whole acapela thing is in flux as we have to completely rework our builds of the voices. Future versions of the voices we sell will only work with NextUp apps and not with other SAPI applications unfortunately. This will consume our acapela work for a while.

[D.Leikin]

I do love using Firefox with TextAloud plug-in, but when I first tried Opera I was amazed how fast, functional, easy-to-use, responsive, and customizable an I-net browser can be. With Opera, web browsing just turns into flying. I had found it nearly impossible to contrive an option that cannot be immediately implemented in just a couple of mouse-clicks. I haven’t been using Opera for a long while, but my fleeting impression is that it’s the cutting-edge solution in web browsing whose functionality can compare to only that of TextAloud in Text-to-Speech conversion.

Any chance you will have a look at the possibility to program a plug-in for Opera? I’d love to see her engaged to TextAloud. IMHO, they both merit being together.

[kdwhite]

We'll put it on the list and see how hard it is.

[SFCurley]

In the pronounciation editor, I would like to see three additional things:

1. troubleshooting/debugging -- more specifically, I would like to see which particular regex is being triggered as sometimes I think it's regex A that's hitting, but it's regex B and I have to go through a lot of effort to figure that out.

2. Order of precedence -- and not be limited to having to a sequential ordering -- for example saying this regex (A) is #1 and that regex (B) is #2 and the remaining regex (C) must, therefore be #3, but instead, have LEVELS of precedence so that regexes A and B can be order of precedence of 1, and C can be order of precedence 2. This would save having to re-order all subsequent regexes when you eventually want to add one higher in the precedence list.

3. A comment field so I can document what a particular regex is supposed to fix since I often forget.

[kdwhite]

Keep them coming. In a week or so we'll review all of this as we build out the wish list.

[BrienMalone]

3. A comment field so I can document what a particular regex is supposed to fix since I often forget.

I was just going to suggest that.

There should also be an extra box in the pronunciation editor where you can type an example sentence to see how your reg ex will behave without affecting the main text window.

[BrienMalone]

Some of these have probably already been suggested, but I figure it can't hurt to have repeats:

Custom Toolbars, Floating & Dockable
Again, I'm looking to MS-Word as the model here. It would be nice to have the standard toolbar up there with New, Open, Save, and Save As... Even Print and Print Preview. Why not? A significant part of TextAloud is a text editor - might as well dress it up like one. Remove and Remove All would be nice on that standard bar as well.

Incidentally, I think the word "Remove" would be better than "Delete"... "Delete" has File Erasure connotations. All you're really doing is removing the article from reading list.

The formatting toolbar would be nice, too. Fonts, font-size, bullets, hanging indent... etc.

Include the ability to make custom toolbars and make them floating & dockable.

Offer the same button display options: small, large, large with text

Word Plug-In
Well heck... Why reinvent the wheel? Why not just make an MS-Word plug-in for TA? I still think the controls mentioned above would be worthwhile, but dropping a TA toolbar in Word would be fantastic.

Text-Edits During Playback
I am always wiping out a word or two by accident just by hitting the keyboard while its reading a passage. I'd like to ask that you either highlight words without using the actual cursor, or provide the option to turn off editing during playback. Optimally, I'd like to be edit independent of the playback, even if TA continues to read the passage as it was when the play first started.

Alt-Keys for Clipboard Watcher options
When TA is watching the clipboard, the "Copy Clipboard to Speech Buffer" dialog appears as soon as text is cut/copied. Most times I'm copying using keyboard commands to copy and paste. If I don't want TA to interrupt, I'd like to be able to stay on the keyboard and [alt]+[a] to append, [alt]+[n] for new or [alt]+[c] or [esc] to cancel. (Those keys are probably taken, but you get the idea)

I'd also like to have an option to decide on the focus behavior after a choice is made on the "Copy Clipboard to Speech Buffer" dialog. The only behavior now is to shift focus to the TA window and the top of the current article, which is appropriate sometimes. More often than not, I'd prefer to have focus returned to the document I cut the text from so I can continue cutting and appending/pasting new.

If I'm appending, (and I've asked TA to send focus to the TA window) it should send me to the bottom of the article that was just appended, not the top.

Also, I've noticed that when the clipboard to speech dialog is up, even though it has focus, for some reason, keyboard commands still affect the previous window. If I'm copying from Internet Explorer, for example, and the dialogue pops up, pressing the space bar will cause the IE window to scroll, even though it is in the background. (I'm pretty sure this wasn't intentional.)

Bullet, m-dash, smart-quote support
It seems like when I paste text with bullets, mdashes and smart quotes, all of the odd characters show up as blocks. I'd like to actually be able to see them.

Paste Special: Unformatted Text
Include a "Paste Special" option, again like MS-Word, that strips out fonts and special characters.

Delete (Remove Article) Confirmation
It would be nice to have a y/n confirmation dialog when the Delete button is clicked.

Perhaps even "This article has changed since it was last saved - Do you wish to save before closing [yes][no][cancel]"

Use the same prompts when exiting the program if the article state has changed.

Font Size/Style/Family identification for RE
I'd love to be able to use RegEx to identify increased font size or underlined text. you could assign different voices to different fonts, speak italicized or underlined words with emphasis... This would open up lots of cool reading ideas.

RegEx and SAPI5 XML to Help File
Please add regular expression and SAPI5 references to the help documentation.

Options
I'm a big fan of customization. If you do implement any of these suggestions, please include the ability to customize or disable the features when practical in the options menu.

[D.Leikin]

A minor option which might be quite helpful is to read aloud subtitles when you watch some DVD (or may be even TV) stuff on PC or PC-powered multimedia centre.

[DaveH]

This may be an idea for V4 but I would like to see the option of voice input to TA via a microphone. A number of my colleagues now use voice input for generating letters and documents but then edit them by hand. It would be nice to have Heather or Kate proof read and then edit the mistakes with voice input. I would also like to use voice input to edit pronunciation and avoid all that fiddling with miss-spelling or phonemes.It would have obvious benefits for the visually impaired and I imagine would open up a whole range of possibilities for further enhancements. An integrated package of this sort could be awesome if technically feasible without costing the earth.

I would also like to see reading bars for WORD and PDF documents allowing one to read documents containing images as with HTML. Adobe reader is quite good but does not highlight the spoken word so that one looses the place relative to the images and it does not allow editing pronunciation (I don't think that the pronunciation edits in TA are used when using TA voices in Adobe reader.....not sure though)

For the more technical reader a scientific version that could read scientific documents (see the posts initiated by 'fedup' concerning OCR) and recognise the range of symbols and expressions available in word would be brilliant. The market might not be large but you would have a monopoly!

I also endorse all of the comments for improving the existing pronunciation editor.

Good luck
Dave.

[geneven]

I just want to chip in, responding to a mention above, that I have recently switched to Opera, the version 9 preview, because I like Opera Widgets. I will probably go back to Firefox sometime, but not soon.

[geneven]

I would like to make a general comment about TextAloud. Perhaps it will help design a new version. I use readers other than TextAloud because they show the graphics of the text I am reading more accurately or pleasingly than TextAloud does. I don't use TextAloud to just LISTEN to text, I LISTEN AND READ at the same time. I find that text looks better in Microsoft Reader. I also like something called Rudenko Reader. And i can't figure out how to read Quicksilver, by Neal Stephenson, using voice, without using Adobe Reader (otherwise I don't like Adobe Reader particularly).

So, any way that you can improve the appearance of text in TextAloud would be welcome. Compare reading text with Microsoft Reader and then TextAloud if you don't know what I mean.

What I DO use TextAloud for is Internet reading, such as news articles.
It's great for a quick translation into speech. I think I would still prefer seeing the text more or less as it was in the newspaper (often the NY Times), with the words highlighted one at a time as TextAloud now does, but with the NY Times display. I understand that this is difficult, but worth mentioning even if not practical.

A really cool additional feature for a reader of newspapers would be the ability to automatically jump to the page continuation (as when it says "more" or "continued on page 5"). Right now, what I do is request that the article be displayed as if it was going to be printed, so it is displayed with no jumps, but more automation would be nice.

(i'm still looking forward to having my paper read to me by Marylin Monroe or John Wayne or some other celebrity. Whoever accomplishes that decently will make a lot of money!)

[BrienMalone]

There should also be an extra box in the pronunciation editor where you can type an example sentence to see how your reg ex will behave without affecting the main text window.

I've been thinking some more about this...

Suggestion for establishing precedence
I have to point to another Microsoft example for setting the order of execution. Microsoft Outlook has the ability to set delivery rules. The rules are executed from top of the list to the bottom. The list is unordered you manipulate individual entries by highlighting one (or a contiguous group) and clicking the "Move Up" or "Move Down" arrows

Handling Different Requirements for Standard vs RegEx Edits
In fact, that whole subsystem could be used as a model. The main dialogue is a list of rules. Here you can create standard, create regex, edit or delete. Create standard and create regex both pop open another 'edit' dialog.

The standard edit dialog just has a "name" for the pronunciation cue, "word" (required) and "pronunciation" (required)... "voice selection", "test", and "save" & "cancel"

Create RegEx has a "name", "filter", "pronunciation", "sample sentence" (defaults to anything highlighted in the body) "voice selection", "test", and "save" & "cancel"

[BrienMalone]

Oooh and another spark from the idea factory.

Scipting Helpers
This next set of request comes from the world of web scripting.

Alternate colors for Mark-up (XML & other) tags

A clickable toolbar that will produce open/close tags for the various mark-up tags.

The ability to highlight a string of text and surround that text with open/close tags

Attribute hints. When a < is typed, use autocomplete hints as the xml command is typed.

When a space is typed after the XML command, pop up the list of attributes for quick select.

When an attribute is selected, if possible, pop-up the range of values.

Optional line numbers

Syntax Validation (open/close tag matching)

All of this type of functionality can be seen with the Adobe (or Macromedia or Allaire) ColdFusion Studio 5.0 when dealing with coldfusion mark-up tags. It's pretty handy.

[geneven]

I would like some kinds of text-handling macros to be part of TextAloud. In my case, when I copy a New York times article, I would like to automatically delete the first few and last few lines of each article, because they have material I have to delete manually now.

[geneven]

When i right-click a link, I would like to have "copy target to TextAloud" be one of the options.

Also, I realized that one of the things I find inconvenient about Textaloud is that it doesn't handle big texts well. That's obviously why you have a File Splitter utility as part of textaloud. It would be nice if you would make it unnecessary to use such a utility. I don't particularly like splitting up my texts. One reason i like Microsoft Reader and some other readers is that I don't have to split up the texts.

[kdwhite]

The right-click link thing, you mean when in IE you'd right-click a link, and the text of that entire web page would become an article in TextAloud?

Describe the issues you have with the way TA currently handles large texts.

[selwyn owen]

I listen to TextAloud-converted books on my walk to and from work and I use the File Splitter utility to break the book down into 21-ish kilobyte files, which convert to 21 to 25 minute mp3 files - just the right amount for my walk - so this utility is very useful to me.

The only improvement to this function is my previously posted request for the splitter to detect paragraph breaks as opposed to any other point (sentence end or mid-sentence punctuation).

Selwyn

[SFCurley]

Another suggestion: a voice-keyed dictionary, where you can specify which voice the pronounciation editor word is to be applied to or a wildcard that means apply to all voices. (This since not all voices mis-pronounce words and since when you enter corrections, voice A can sometimes pronounce it one way, while voice B might pronounce it a second way.)

[D.Leikin]

If possible, please have a look into the possibility to program a plug-in for OpenOffice Writer.

Here are some reasons why this might be worth doing.

OpenOffice Writer is a multiplatform, multilingual advanced text editor. In contrast to MS Word, OpenOffice Writer is an open-source and lends itself to programming in a variety of programming languages.

OpenOffice Writer is a fully featured word processor of much the same functionality as MS Word and it has nearly the same interface as Word does. (It’s important that MS docs with graphics, tables, and sophisticated formatting will look in Writer quite the same as in Word.)

Writer can open files in DOC, RTF, and many other popular formats and what’s more it can save files in MS DOC and RTF formats, export documents to web in HTML, publish in PDF, and etc.

TextAloud + Writer could become a first text editor with extended text-to-speech capabilities or alternatively a sui generis Text-to-Speech application having the functionality of an advanced text editor.

Writer is free for use and redistribution.

(Licensing issues are addressed in details at OpenOffice.org and related sites.)

[geneven]

If I take a large book and copy it to TextAloud without splitting it, the reading becomes hurky-jerky. This doesn't happen much with other readers. I assume that it's because larger files use a lot of memory or other resource. Luckily, most other readers don't mind as much.

You mean you didn't create the File Splitter program because of problems users have with large files, but just because of benevolent wishes for users who have large files and would prefer smaller ones? I assumed it was the former.