Info

You are currently browsing the archives for the voice category.

November 2008
M T W T F S S
« Oct    
 12
3456789
10111213141516
17181920212223
24252627282930

Archive for the voice Category

Xhtml+voice Part 2

I’ve tried uploading a version (not very much different to yesterday’s) to Voxeo and testing through Skype but I keep getting “an internal error has occurred”. The voicexml appears to be valid. The xhtml is valid, I’m still not 100% sure about the JavaScript. You can’t check the whole thing on the regular w3 validators. I am too tired to process any more code tonight.

In terms of saving something as a favourite, via a cookie downloaded to the device, it appears to depend on different devices and manufacturers, session cookies appear to work across all, but persistent cookies may disappear once phone is switched off.

In theory, it should be possible to create a series of voicexml forms referenced in an xhtml+voice page which allow someone to choose what type of web coding template they would like to receive and then when the final form is submitted, a mouseclick or spoken event should send a message to the server to deliver back the output of those choices.

Its similar in principle to any type of online shopping. Once you have completed payment an invoice is generated - either on the final ‘thank you’ page or a ‘confirmation that it has been emailed to you’ page. As with a milkshake order, it should be possible to return a receipt of your order as a text file after you have completed the order, so sending a text file back to the user with an xhtml, RSS, php or whatever type of coding template inside it, which can be then saved to the phone or device should be no different.

A guess of how to do it would be to put templates in a mysql database then add php to the original xhtml+voice page to tell the server to fetch the template and include it in an email/SMS back to the customer. I know php has code for email options and for attaching files, I just don’t know exactly how to do it myself…yet - anyone else out there using or developing an ecommerce site tried ?

If this is all possible, it also means that someone who wanted a template might not have to actually go online at all, they could just make a phone call, make their choice and receive the template back - i.e. they would be calling (just like telebanking). Hopefully….

Will return to this later - am SO not on track to complete everything by September (maybe end of) but its still been very useful exercise, uncovering increasingly complex layers of coding, media and to some extent - architecture, now starting to make some connections between some of these too.

There is definitely potential to create pages of code against the 5 original strands (sms,multimodal,standalone vs mobile web/widget, pc vs web, mobile opensource vs mobile proprietary) it is possible to create both stand alone and web applications to do this, will compare and review their efficiency against the other strands in terms of device independence in the final stages of this project. Will also compare and review as to what this could mean for any mobile learning development and if it is worthwhile being deployed.

Anyway, my online persona is going to have a break for a few days, not sure what the offline one is going to be doing yet :-)

erm, slowly weaving an xhtml+voice document (this is only half a post really)

First unfinished attempt which have not either validated or checked via Voxeo. Please feel free to copy / download (will need to select page and source to view code….once I have finished version will upload as a text file)
Would also recommend waiting until have a finished attempt that is assuming anyone out there would actually want a copy of either finished or unfinished ;-) It will hopefully provide at least half an idea. The milkshake one is more complicated in that you need branching immediately upon answering the first question. I guess it is unlikely that if you had such a wide choice of flavours that you would just tell an online form that you wanted one flavour only.

Is based around this text from post yesterday.

It is quite similar to a familiar Xhtml page except that it has slightly different DOCTYPE and namespaces to be declared. Another difference is use of time - which you can declare in seconds or milliseconds, you also should provide script so that the browser reading it, knows that you’re finished / stops listening. You also use scripting (JavaScript) to manage events such as prompts, speech input and branching etc I’m not entirely sure I’ve got my “elements by ID” in the right sequence to fit with the branching at the moment, but will inspect in more detail tomorrow

You can put the voicexml directly into the Xhtml coding or reference it as an external voicexml file (based on everything I’ve understood so far, external is better because you can reuse it, even within the same document). You set up grammar files to help with understanding i.e. a file will include typical words that you might expect to hear in response to a prompt/question.

Tomorrow’s task is to reference this unfinished version against the W3 guidelines Then put together an aural stylesheet. Another option is to record the prompts and save as audio files which you can then reference within the forms, so might do this too.

First thoughts about bringing in multimodal web and Voicexml / Xhtml+Voice

Mentioned this on the original post about this mini-project

“…multimodal web & messaging only, so I understand it as….a future extension of mobile web and learning - giving people choice over how they interact and use information and for the purposes of this little project, looking at voice-text only. A multimodal browser will allow multiple types of input or output using a range of options - keyboard, touch, voice etc I will be looking at Xhtml + Voicexml (x+v) markup languages”

Have deliberately not done much with due to other more intense programming bits but thought it was about time to post something. Prior to May, had not attempted to write/code anything related to voice before, although wrote an audio stylesheet about 2 /3 years ago.

Trying to think this along the lines of:

1. Should it be a mobile web page with a basic form ? This would not appear to be what xhtml+voice forms have already been designed for (e.g. examples so far seem to be things like using a form to find something or order pizza etc), but was wondering whether you could create a basic mobile web page, with options when selected by voice input would return a web page template as a file, to the user either via SMS or the web. If you’re going to be online anyway, would using voice add any value in terms of speeding up the process, so you could edit / create mobile web pages on the go ? Hmmm, not sure, but having both a visual and voice form would make it more accessible.

2.If not a whole mobile web page, could it be a regular web page with voice annotated ‘elements’ on the page that could produce a similar result ? Would it be better than a mobile web form ? Don’t know yet until try.

3. Due to limits of connectivity / bandwidth / cost / time, would it be more useful just to provide a standalone mobile voice application that could produce a set of templates and is it possible ? Could any files or the code be generated by a form similar to the above to generate either a voice or text SMS? (As above, re using voiceSMS - from an accessibility point of view - is better to have both a text and voice option?) Also don’t know yet until try. (Will return to this in a future post).

Will not be starting with an example of a mobile-friendly web page just yet because as per mobile web experimenting, haven’t done that yet - however will be bringing into this project later on.

In order to accept user input i.e. someone saying something, this is coded on a web page in a similar way to inputting text, using a form. So just as each text box would appear on a page, there can be an equivalent set of code to allow for speech. The examples provided by others would be things like - ordering food or drink, so if you were using your mobile phone and wanted to quickly order something like a milkshake (why you would be ordering one via the web, but off the top of my head…) your order would be a set of speech ‘events’.

So, am wandering around in Bournemouth, UK, its a lovely sunny day and I want a milkshake - so decide to order from Shakeaway (if you are ever in Bournemouth….) so a form would include questions like

Would you like a milkshake with:
Chocolate bar
Cake
Biscuits
Cereal
etc

Would you like anything else with it?
(note - if it was Shakeaway then a mobile web form would be totally inappropriate because there are hundreds of combinations)
Strawberry
Cherry
Lime
Kiwi
Melon
etc

Dairy or alternative?
Dairy
Soya

Any extras?
No thanks
Extra icecream
bran
protein
etc

Finished?
Yes
No

Would you like to save this as a favourite?
Yes
No

All of the above could also branch out - imagine if you were standing at the counter and being asked what you wanted, with lots of choices to look at, you might need several decisions or you might have a favourite - mine is chocolate & cherry with soya - so on the form I would be saying chocolate then selecting cherry then soya, no thanks then yes to finish. However if you would like anything else (question 2) you could potentially have additional sub branches e.g. 2a being lists of chocolate bars, 2b being cake options etc etc, depending on how many flavours you wanted.

In voicexml forms, each question would be treated as its own ‘block’ so you would need to add in extra code to tell the form to go to other blocks. No idea exactly at this point how to code the cookie bit (i.e. would you like to save as a favourite, am assuming that you can save cookies on a form on mobile device in same way as pc, but haven’t looked into as yet), but it would be useful seeing as you are using a phone and want to save time being online.

What does this look like in voicexml?
Will try and post tomorrow. Am hoping that same principle of menu items would apply for this project i.e.
What kind of page do you want to create?
xhtml
RSS
php
etc

Then a branch if selected Xhtml:
What type of Xhtml?
Strict
Transitional
etc

or if selected RSS -
Is this for a podcast?
Yes
No

(If select yes, this would be telling the server that you want RSS 2.0 and automatically bring in a template with itunes, encl and other media-specific tags)

This is about as far as I have got in terms of thinking it through. There are a whole bunch of resources available and I spent (productively going round in circles) several hours trying to work out how it was possible to test anything without having access to your own voice server. I have signed up to Voxeo because you can test anything you create using Skype which sounds useful. You can also enable voice if you download Opera browser (Voice is not on Opera mini yet.)

Refs and sources for this post on delicious with voicexml tag

|