VoiceXML 2.1 Development Guide

| TOC

w3c Properties

The available VoiceXML browsers offer support for most, but not all, of the properties listed in the w3c 2.1 specification. The complete list of supported properties is detailed below:

Property	Value
audiofetchhint	`safe\|prefetch`

Details: The audiofetchhint property designates when audio content should be fetched from the server. When set to 'safe', (default), then the audio is only fetched when needed. When set to 'prefetch', then the platform will attempt to pre-load the audio files to optimize performance.

Property	Value
audiomaxage	`CDATA (86400\|one day default value)`

Details: The 'audiomaxage' property informs the platform exactly how long, in seconds, the maximum acceptable date range is to cache audio content. Note that cache control is always handled best by server-defined settings.

Property	Value
audiomaxstale	`CDATA (3600\|one hour default value)`

Details: The 'audiomaxstale' property defines how long, in seconds, that the maximum acceptable staleness of audio content should be. Note that cache control is always handled best by server-defined settings.

Property	Value
bargein	`true\|false`

Details: The 'bargein' property controls whether or not the TTS or audio prompts may be interrupted by a caller's voice input. This property defaults to 'true', unless specified.

Property	Value
bargeintype	`speech\|hotword`

Details: This property allows the developer to define which type of bargein, (via speech, DTMF or 'hotword' input), is to be considered a valid method of input to the application.

Property	Value
completetimeout	`CDATA`

Details: The 'completetimeout' property specifies the length of time after a user utterance where the interpreter will consider the utterance complete. Changing the default value of '.25s' upwards offers more reliability in catching a user's complete utterance, at the cost of sacrificing response time. This property will only be active when set to a lesser value than the incompletetimeout property.

Property	Value
confidencelevel	`CDATA`

Details: The 'confidencelevel' property allows you to adjust how accurate a user's utterance needs to be to get a valid grammar match. If one were to change the default value of '.45' to '0.1', then any utterance would be matched, while if it were to be adjusted to '1.0', then virtually every user utterance would be considered a 'nomatch'.

Property	Value
documentfetchhint	`prefetch\|safe`

Details: This property setting defines the behavior for document fetching within the application. When set to 'safe', then documents will be fetched as-needed. When set to 'prefetch' (default), then the content will be pre-fetched, (if possible) to speed up fetch times.

Property	Value
documentmaxage	`CDATA (0s default)`

Details: The 'documentmaxage' setting allows the developer to specify how long, in seconds, the maximum acceptable date range is to cache the document(s) in question. Note that cache control is always handled best by server-defined settings.

Property	Value
documentmaxstale	`CDATA (0s default)`

Details: The 'documentmaxstale' property defines how long, in seconds, that the maximum acceptable staleness of application content should be. Note that cache control is always handled best by server-defined settings.

Property	Value
fetchaudio	`CDATA`

Details: The 'fetchaudio' attribute specifies the URI of a .wav file to play to the caller in the event of an extended document fetch. Essentially, while the fetch is being made, it allows the developer to play some filler music to the caller rather than presenting only silence.

Property	Value
fetchaudiodelay	`CDATA (2s default)`

Details: This property setting designates how long to wait before playing the wav file specified in any 'fetchaudio' attributes/properies. In some cases, such as a comparatively short, (2-4s), document fetch, it is 'smoother' to present silence rather than 1 second of the fetchaudio 'filler'.

Property	Value
fetchaudiominimum	`CDATA (4s default)`

Details: This property defines the minimum amount of time that any fetchaudio content is allowed to play. Often used in conjunction with the 'fetchaudiodelay' setting to refine fetchaudio behavior on a Global scale.

Property	Value
fetchtimeout	`CDATA`

Details: The 'fetchtimeout' property allows the developer to specify how long the Voice Browser should attempt to fetch a document before throwing a fatal 'error.badfetch' event.

Property	Value
grammarfetchhint	`safe\|prefetch`

Details: This property specifies when external grammar files should be fetched. The value of 'prefetch', (default) means that grammars are fetched whenever the VoiceXML page itself loads, whereas a value of 'safe' will only fetch the grammar when it is called for.

Property	Value
grammarmaxage	`CDATA (0s default)`

Details: The 'grammarmaxage' setting allows the developer to specify how long, in seconds, the maximum acceptable date range is to cache the grammar(s) in question. Note that cache control is always handled best by server-defined settings.

Property	Value
grammarmaxstale	`CDATA (0s default)`

Details: The 'grammarmaxstale' property defines how long, in seconds, that the maximum acceptable staleness of grammar should be. Note that cache control is always handled best by server-defined settings.

Property	Value
objectfetchhint	`CDATA`

Details: Tells the platform whether the URI contents for <object> may be pre-fetched or not. The values are prefetch (the default), or safe.

Property	Value
objectmaxage	`CDATA`

Details: Tells the platform the maximum acceptable age, in seconds, of cached objects. The default is platform-specific.

Property	Value
objectmaxstale	`CDATA`

Details: Tells the platform the maximum acceptable staleness, in seconds, of expired cached objects. The default is platform-specific.

Property	Value
incompletetimeout	`CDATA`

Details: The 'incompletetimeout' property allows you to change the setting of when, after a caller pauses in mid-utterance, the interpreter will start to process said utterance. After the period of time specified, (2s default), the interpreter will process the utterance and return either a valid grammar match, or a 'nomatch' condition.

Property	Value
inputmodes	`dtmf\|voice`

Details: This property allows the developer to specify what types of input ()voice or dtmf) are allowable within the application. When set to 'dtmf' the application will ignore all voice input, and vice-versa.

Property	Value
interdigittimeout	`CDATA`

Details: This property specifies the amount of time between multiple dtmf inputs that a caller is allowed before the interpreter decides that the caller has finished. To tune any appreciable delay between dtmf input and the recognition return, the default value of '4s' can be adjusted downward to a more comfortable value, although at the expense of possibly truncating the caller's full dtmf input.

Property	Value
maxnbest	`CDATA`

Details: This property specifies the maximum number of interpretations that should be returned from the recognition when NBest post-processing is performed in the application. The default value for this property is '1'.

Property	Value
maxspeechtimeout	`CDATA`

Details: This setting defines the maximum duration of a user's speech that is to be accepted before throwing a 'maxspeechtimeout' event.

Property	Value
recordutterance	`true\|false`

Details: This property gives us the entire utterance and places it into the variable application.lastresult$.recording.

Property	Value
recordutterancetype	`audio/wav\|audio/x-wav\|audio/basic\|audio/x-alaw-basic`

Details: This property allows the developer to define the resulting audio format of the utterance recording. Details on the audio file types are listed below:

audio/wav -- (default) container format = .WAV | meta-format = RIFF | compression = ulaw | 8bit, 8kHz, 64kbit/s, mono
audio/x-wav -- container format = .WAV | meta-format = RIFF | compression = ulaw | 8bit, 8kHz, 64kbit/s, mono
audio/basic -- container format = .WAV | meta-format = RAW (headerless) | compression = ulaw | 8bit, 8kHz, 64kbit/s, mono
audio/x-alaw-basic -- container format = .WAV | meta-format = RAW (headerless) | compression = alaw | 8bit, 8kHz, 64kbit/s, mono

Property	Value
scriptfetchhint	`safe\|prefetch`

Details: This property specifies when external script files should be fetched. The value of 'prefetch', (default) means that scripts are fetched whenever the VoiceXML page itself loads, whereas a value of 'safe' will only fetch the grammar when it is called for.

Property	Value
scriptmaxage	`CDATA (0s default)`

Details: The 'scriptmaxage' setting allows the developer to specify how long, in seconds, the maximum acceptable date range is to cache the script(s) in question. Note that cache control is always handled best by server-defined settings.

Property	Value
scriptmaxstale	`CDATA (0s default)`

Details: Tells the platform the maximum acceptable staleness, in seconds, of expired cached scripts. The default is platform-specific.

Property	Value
sensitivity	`CDATA`

Details: The 'sensitivity' property allows you to adjust how sensitive the interpreter is to loud versus quiet input. Adjusting the default value of '0.5' upwards will make the interpeter more sensitive to quiet input, while adjusting this value downwards will make it less sensitive to quiet input. This property can be useful if your end user will be accessing the application in a very noisy environment, although it is advised that you use this property very carefully, as it may have a detrimental effect on recognition accuracy.

Property	Value
speedvsaccuracy	`CDATA`

Details: The 'speedvsaccuracy' property allows you to tune the trade off between utterance matching accuracy, and the speed of the return. If the default value of '0.5' is adjusted upwards, accuracy increases, but the speed of the return is hampered. Adjusting the value downwards will return results faster, at the expense of accuracy. Use this property with extraordinary caution, as this can negatively impact your application if used carelessly.

Property	Value
termchar	`(any dtmf key)`

Details: The property 'termchar' allows you to set the default dtmf key that signals the interpreter that dtmf entry is complete. This value, unless specified, defaults to the pound key '#'.

Property	Value
termtimeout	`CDATA`

Details: The 'termtimeout' property specifies the terminating timeout to use when recognizing DTMF input.

Property	Value
timeout	`CDATA`

Details: The 'timeout' property allows to to specify how long the interpreter will wait for a user utterance before it will throw a 'noinput' event. This property defaults to '5s', unless explicitly specified.

Property	Value
universals	`all\|none\|help\|cancel\|exit\|repeat`

Details:This property allows you to disable or enable the universal grammars for your application. By default, the value for this property is set to 'none'. See the section on universal grammars for further details.

ANNOTATIONS: EXISTING POSTS

awirtz

11/29/2004 3:10 PM (EST)

The values assigned to these properties should not be enclosed in single quotes like values assigned using the assign tag. (since this is a literal CDATA value, not a JavaScript expression)

Also note that when specifying a duration in seconds, the "s" is indeed part of the literal CDATA value.

Example:
<property name="incompletetimeout" value="4s"/>

safarishane

2/9/2006 12:12 PM (EST)

Is the W3C property for recording utterances during asr reco supported?

<property name="recordutterance" value="true"/>

http://www.w3.org/TR/voicexml21/#sec-reco_reco

Thanks,
Shane

steve.sax

2/9/2006 12:40 PM (EST)

Hello Shane,

At present, this 2.1-specific property is not enabled on out VoiceXML platform, but we do have plans for enabling this on our Prophecy platform in the very near future.

However, you can achive the very same functionality at present by employing the <voxeo:recordcall> extension element. Check the 'elements' section in the left-hand navigation pane to learn more about this element, and it's usage.

Warm Regards,

~Steve Sax

safarishane

2/9/2006 12:54 PM (EST)

Actually, I'm looking to only record the <nomatch> portions of a call. Please let me know when Prophesy is available in your hosted environment.

Thanks

steve.sax

2/9/2006 12:56 PM (EST)

hello Sahne,

Prophecy is available in the hosted environment already, but we don't yet support the property that you mentioned. bear in mind that Prophecy is still a Beta offering, so we can expect changes and updates daily.

~Steve

beta

8/23/2006 6:14 AM (EDT)

Isn't it included with Prophecy 7 now?

MattHenry

8/23/2006 1:02 PM (EDT)

Hi there,

Getting shadow variable results in the context of a nomatch event is a feature that we have not yet deployed to either voice center offering. This particular feature will require some sizable engineering effort to complete, and while this feature is on our to-do list, we do not yet have a date of deployment finalized.

When this is enabled, we will be sure to post details in our release notes, available in the Prophecy forums.

~Matthew henry

esirkin

10/11/2006 3:54 AM (EDT)

<property name="bargeintype" value="dtmf"/>

is not permitted according to the W3C spec. The spec says that only "hotword" and "speech" are to be used. The spec then goes on to say that using "speech" enables "dtmf", one should set

<property name="inputmodes" value="dtmf"/>

to restrict inputs to dtmf tones.

If this is something specific to Voxeo platform, then it is a nice feature but shouldn't it be documented as such?

eric

MattHenry

10/11/2006 11:27 AM (EDT)

Hi Eric,

This is a case of documentation meant for our older VXML platform, (Nuance VWS 2.0), seeping into our docset for the Voicecenter 5.5/7.0 documentation. Note that this property setting has no effect on the contemporary platforms we offer; the setting will just be ignored.

All the same, I'll see about getting this corrected in the docs so that this doesn't confuse anyone.

~Matthew henry

fliptone

8/5/2007 3:09 PM (EDT)

above you say "Note that cache control is always handled best by server-defined settings". Where are these settings defined?

VoxeoBrian

8/5/2007 3:19 PM (EDT)

Hello,

What is meant by setting cache control on the server side is specific to the web server you are using locally to server up content.

Here is an excellent link that outlines setting HTTP cache control headers, as well you should be able to find documentation specific to your server program, ie, Apache, IIS, etc.

http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html

Hope this helps!

Regards

Brian

mtatum111

4/23/2009 10:51 AM (EDT)

Is there anyway that you can have two values for termchar?

I know that the default is "#". However, what if you want it to also allow "*" as a termination key. I don't think that you can have two according to the spec, but just want to make sure. I know that you can use the <property> element and set the termchar before certain prompts. However, that is still allowing on one value for the termination character.

Thanks for any thoughts.

mikethompson

4/23/2009 11:25 AM (EDT)

Hello,

I've checked the W3C VXML spec, as well as conferred with our VXML Platform Engineers, and it was verified you can only specify a single termchar via the termchar property. Of course, you can always create your own grammars with the termchars built in to them. For example this could be part of your SRGS grammar once you're done collecting the other input:

[code]
<one-of>
<item>#</item>
<item>*</item>
</one-of>
[/code]

Hope this helps,
Mike Thompson
Voxeo Corporation

jvincent

4/14/2010 11:43 AM (EDT)

It would be great if this page had anchor tags for each property. That way, I can put hyperlinks in my administrative web-app directly to the property for which my user is configuring the tool to use.

Extremely useful resource otherwise. THANKS!

voxeoblehn

4/14/2010 3:11 PM (EDT)

Hello,

Thanks for your feedback! Our documentation is constantly evolving and we will certainly take this into consideration in future revisions of our docs.

Best Regards,

Brian Lehnen
Voxeo Support

poonam007

5/14/2010 7:20 AM (EDT)

Hi,
what is the significance of speedvsaccuracy property?and how are the completetimeout and incompletetimeout properties to be used?

Thanks,
Poonam.

voxeoJeffK

5/14/2010 7:44 AM (EDT)

Hello,

The speedvsaccuracy property just controls how much time the ASR engine should spend in determining match. It uncommon to need to change it. One example would be for when a larger grammar takes a long time to return a match, and you want it to return faster to avoid making the user wait. The downside is you will reduce the accuracy of the matching.

Concerning 'completetimeout' and 'incompletetimeout'. 'completetimeout' property is set, by default, to only 250 milliseconds. It is a measure of the amount of time the interpreter will wait after an utterance to consider it complete and continue with execution. In a case where your grammar entries are single words, if the utterance has matched one of your single-word entries, and there are no multiple-word grammar entries that could be a possible match, there is no reason for the interpreter to continue waiting for more input. The utterance is considered complete, and processed after just 250ms. If there is indeed a multiple-word grammar entry that is a possible match, the amount of time the interpreter would wait for the additional input is controlled by the 'incompletetimeout' property, which is set to the default value of 2 seconds. Again, if there is not a possible multi-word match, the interpreter will not wait 2s ('incompletetimeout') for additional user input, it will simply process the utterance / single-word match in only 250ms ('completetimeout') after the end-user says the matched word. So adjusting these settings depends on your type of grammar as well as how flexible the recognizer is for users with - long - delays - between - their - words.

hope that helps,
Jeff Kustermann
Voxeo Support

poonam007

5/15/2010 11:07 AM (EDT)

Thanks Jeff,
Your explanation has made it pretty clear...but i just wanted to confirm one point...that is are completetimeout and incompletetimeout both effective for a grammar file containing single word as well as multi-word entries?

Regards,
Poonam.

VoxeoBrian

5/15/2010 3:26 PM (EDT)

Poonam,

Good question, in the scenario of single word grammars such as a grammar that would accept yes or no, incompletetimeout really does not come into play since this is associated with the pause between utterances, such as what you would expect user repeating back digits, ie: 1-->2-->3...

Completetimeout would be really the deciding factor for single utterance grammars.

Hope this helps!

Brian F.

poonam007

5/17/2010 4:38 AM (EDT)

Thanks Brian,
You have made the ussage of completetimeout more clear..but i still have one doubt...
If i have a grammar file containing single word and double word entries for example like the following:
<one-of>
<item>coffee</item>
<item>tea</item>
<item>coke</item>
<item>lemon tea</item>
<item>lemon juice</item>
</one-of>
and completetimeout="250ms" and incompletetimeout="2s" ,then like what Jeff has said will happen "If there is indeed a multiple-word grammar entry(lemon tea) that is a possible match(in case the user has said only lemon ), the amount of time the interpreter would wait for the additional input is controlled by the 'incompletetimeout' property, which is set to the default value of 2 seconds. Again, if there is not a possible multi-word(coffee) match(in case the user has said coffee), the interpreter will not wait 2s ('incompletetimeout') for additional user input, it will simply process the utterance / single-word match in only 250ms ('completetimeout') after the end-user says the matched word."
So here both completetimeout and incompletetimeout are effective.
And in a scenario where a grammar file contains only multi-word entries,completetimeout property would be ineffective.
kindly let me know if i'm right.

Regards,
Poonam.

voxeoJeffK

5/18/2010 1:37 AM (EDT)

Hello Poonam,

Yes, your understanding is correct.

Regards,
Jeff Kustermann
Voxeo Support

H: Property Guide

| TOC

| Nuance ASR Property settings