Monday, May 27, 2024

Selecting a VoiceXML Gateway

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

By Jonathan Eisenzopf

A VoiceXML gateway is the server software that handles
interactions between callers on standard analog lines (or VoIP) and
voice applications on a Web server.

In a way, the fact that a common
standard exists for IVR functionality means that VoiceXML gateways
could soon become a commodity technology offered by every phone device

Until the time when VoiceXML functionality comes
standard with your PBX or phone system, you may be stuck with the
difficult task of selecting an independent VoiceXML vendor.

Vendor Standardization

If your company has standardized on a particular vendor for
networking or telephony gear, your best bet is to contact the vendor
to see if they offer a VoiceXML gateway product. If they do, chances are
that it will be easier to integrate the vendor’s product with your other equipment.
Cisco, Lucent, Nortel, and Siemens are all examples of companies who
can provide a VoiceXML gateway that integrates with their telephony
and/or networking equipment.

Have your requirements in hand

Don’t make the mistake of letting the vendor dictate your
selection criteria. This happens when you start talking to vendors
before you have determined what your needs actually are. It’s very
easy to fall into the trap of comparing vendors based on a
differentiator that one vendor emphasizes when it may not be of any
value to your company.

Before you approach the first vendor, make sure you have the
following information at hand:

  • Expected call volume – How many calls do you expect per day?
    How many simultaneous calls do you expect at peak volumes?
  • Legacy system – What equipment will the VoiceXML gateway need
    to integrate with? Common equipment that you need to be
    concerned with would be your PBX, ACD, phone system, or IVR
  • TTS channels – Will your system require any text-to-speech
    functionality, or will every prompt be be-recorded?

Deciding Factors: Cost, Support, Ease of Integration, and
Development Tools

Once you have narrowed your list of potential vendors, your
decision will likely be based on cost, the quality and level of
customer support, the level of complexity required to integrate the
solution with your existing infrastructure, and the development
tools that come with the VoiceXML gateway.

Most vendors use the same underlying hardware and software, so
performance, scalability and reliability of all platforms will be
comparable with some exceptions. 

One area where vendors differ is in the way they extend the
platform to integrate with other systems. For example, one vendor
may offer seamless integration with Java application servers, while
another focuses on packing high-performance into a single connected
platform. Another vendor may focus on integrating with automated
outbound call systems or call distributors. Some vendors support
Windows, some Unix, some both. Based on your specific integration
needs, you should be able to narrow the number of qualified vendors
down to 2 or 3.

Another area you should focus on is development tools. Each
vendor provides their own unique set of development tools that are
tuned to work with their platform. Have one of your developers work
with the tool to ensure that it meets your expectations and needs.
Because VoiceXML itself is a fairly new technology, not all of the
development tools are fully mature, so be careful.

Then ask yourself what you will expect from the vendor in terms
of technical support. Are they ready to provide the needed level of

Port density

Port density refers to the number of telephone connections that a VoiceXML gateway can handle on a single server.

Most VoiceXML gateways will scale to meet the needs of large call
centers, however, very large installations may require special
hardware. Most platforms are capable of handling up to 4 T1
connections or 96 telephone connections (24 per T1).

Practically speaking, if you require Text-To-Speech (TTS) and
Automatic Speech Recognition (ASR) functionality, a single server
should not exceed 24 ports (or a single T1 telephony board). The
reason for this is that each TTS and ASR channel can utilize a great
deal of resources. If we were to divide 1 gigabyte of memory by 24
ports, that would give us approximately 42 megabytes per port, which
may not be enough to run a TTS and ASR channel simultaneously. CPU
resources are an even bigger issue. 

Typically, you will vary the number of TTS channels based on the
expected usage, since most applications will use pre-recorded
prompts and limit the use of the TTS engine. A reasonable ceiling
for a VoiceXML gateway is around 8.

When using port density as a basis for comparison, you must keep
these issues in mind. A recommended course of action would be to
determine the number of TTS and ASR channels that will be required
along with your expected call volume.


Every vendor will claim that their platform performs well, but
does it really? You should confirm performance claims through
independent test analysis. CT Labs is one such firm that provides
testing services for evaluation purposes. When performance becomes
an issue, the usual course of action is to run the TTS engines on a
separate server from the ASR servers. When the two are segmented,
the ASR server should be able to handle up to 96 voice ports. The
maximum number will depend on the performance of the vendor’s
VoiceXML interpreter.


Find out what the procedure is for upgrading the platform when
more capacity is needed. Will you have to rebuild the server? Can
you chain other servers together and have them work in parallel or
will you have to segment the servers and manage them independently?
Find out the maximum number of ports the platform can handle at

There are two basic ways to scale the platform. The first is to
add additional TTS, ASR, or media gateway servers based on demand.
These systems communication over a network.

The second way is to use a passive back plane where processor and
telephony boards slide into the chassis as needed. A back plane
configuration will likely perform better than a networked
environment because of the inherent latency of network
communications versus a shared hardware bus.

Financial Stability

Because most VoiceXML gateway vendors are relatively new
companies, their financial viability should be a concern. If you
aren’t confident that the company is stable, ask them to reveal
their financial statements so that you can be comfortable that
they’ll be around to service the platform in the future.

Customer Support

Look into what support is available from the vendor after you buy
the gateway. Will they help you set it up? Do they have a
professional services team that can help you integrate the platform
with your enterprise? Keep in mind that some vendors are more
focused on developing the product and would prefer to rely on
partners for customer support and services. In these cases, make
sure you know who will be supporting the product and make sure
you’re comfortable with the third party if the vendor will be
referring you elsewhere for support and integration services.

List of VoiceXML Gateway Providers

Jonathan is a member of the Ferrum Group, LLC  which specializes in Voice Web consulting and training.

This article was first published on, an site.

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles