You are on page 1of 5

BUSINESS MADE SIMPLE

Product Brief
Why Text-to-Speech?
The robots are gone, and amazingly natural-sounding speech has opened a whole
new horizon.
Introduction appreciation for their business and the dramatically slash the costs of their
message that you are trying to build a voice-driven applications, add incredible
How we speak to each other says a lot loyal and supportive relationship? The flexibility and speed to market of
about a relationship. You can tell if most demanding firms go to great dynamic services, and still control the
someone is angry or pleased by his or lengths to ensure that every customer branding and company persona. The
her tone. You can get a different touch point has a unified look, feel and robot-sounding voices are truly gone
message from what the words are saying personality that best projects the with the latest TTS solutions, and the
sometimes by the intonation or pitch of corporate brand and message. But, with amazingly natural-sounding speech
the communication. Change the volume such pressure on expenses, how do you output has opened a whole new horizon
of certain words, and again the meaning think these top performers continue to for automated solutions.
can be altered. The result that is deliver the numbers and still dazzle
delivered is dramatically different than customers with incredible service Why Text-to-Speech
what was intended or is right on target. delivery? One of the secrets is today's
Text-to-Speech (TTS) technology that There are many situations when your
How do you communicate with your has achieved a level of quality and customers prefer a self-service method
customers? Do they get your sense of natural tone that enables companies to of retrieving information, but certain
data is difficult to record. There might
be too much data. What if the database
is extremely large, for example, a
product listing with 50,000 entries?
Regardless of size, the database may be
too dynamic to maintain in a human-
recorded manner. If the data changes
daily or weekly, the company may not
want to spend the time and money
rehiring the voice talent or talents,
renting the studio and technicians and
so on, every week or every month. Text-
to-Speech is the answer.
Converting Text into Ordinary Although this technique has been used prosody, speed, pitch, volume voice and
Speech with some success, typically in niche language. This high quality approach has
applications, most perceive this type of led the major TTS engine providers to
Text-to-Speech (TTS)—or speech Text-to-Speech engine as sounding offer a wide variety of languages as well
synthesis technology, as it is sometimes robotic. as male and female voices.
referred to has advanced a great deal in
just the last few years. TTS converts A Turbo-Charged Engine User Control to set the Mood
ordinary ASCII text or other textual
information into intelligible speech that The newer, more advanced TTS engines TTS can be a complex technology to
now closely resembles a natural voice. It concatenate the speech synthesis master. The latest engines now include
brings considerable advantages over the approach. In this procedure, waveforms extensive user controls that can be
prerecording of prompts from a time, are reclaimed from a database of pre- utilized to enhance the quality of speech
cost and storage capacity requirement recorded speech. The output sounds output generated. Application
perspective. Pre-recorded speech quite natural because the delivered developers can massage raw text strings
requires strict limits to what will be speech is essentially made up of by embedding special control characters,
spoken based on a defined script. Often, concatenated units of human known as tags, to convey emotional
the larger the script, the greater the cost recordings. The resulting speech qualities in the speech output. Phrases
to record and time needed to prepare. delivered is of immense quality and can be adjusted by speed, pitch, volume,
Which means losing valuable time-to- completely natural that some users pace and inflection to convey a
market, or worse, time-to-money cannot tell the difference from recorded particular mood, urgency and/or clarity
opportunity. TTS communicates human interaction. In addition to quality required. This feature is one more
information to customers when possible improvements, the newer engines element that has supported the great
selections include large numbers of require less storage space than pre- advance in TTS.
items from databases that must be recorded speech and can read back any
spoken to callers, or when a list of word, phrase or sentence. If you have
selections changes regularly. For used an Interactive Voice Response
example, this technology would be (IVR) system to secure information that
effective for an organization that needs deploys TTS embedded in the
to confirm street addresses of callers application, the transition from “Enterprises with
and whose database contains many recorded information to speech
telephone applications
thousands of different addresses, or for synthesis has been choppy and obvious.
the reading of email, news alerts or Newer solutions have found a way of
requiring the delivery
other dynamic content. creating a seamless transition between of a wide range of
the prompts and text-based output to information to a caller
The Early Days greatly improve the sound of the should evaluate the
application and the user experience.
During early attempts at synthetic Improving the quality of a TTS
latest TTS technology,
speech, memory was extremely application is a rather technical and as the quality is
expensive. This expense affected the complex undertaking. It requires significantly higher
way early synthesis engines were knowledge of what to say, known as than the previous
designed. A memory-efficient synthesis phonetics, and how to say it, known as generations.”
by rule system was popular, most prosody. The end result – and its
commonly known as a formant engine. positive acceptance by your customers –
—Steve Cramoysan, Gartner
A formant synthesizer created totally come from the careful modeling of the
digitized or synthetic speech with no phonetics and prosody concepts. Many
human recordings used. One advantage new TTS engines now support the
of this method is that the pitch and W3C's VoiceXML and SSML
duration of words could be varied. specifications, allowing application
However, the sound quality is inferior. developers to tune properties such as
There is Always an Exception The Cost Saving Implications Certain engines may support one or
multiple methods. Having an expert
An exception dictionary is critical, since Using real-time speech synthesis that model the variation can have a major
it can define the pronunciation of sounds incredibly like a human voice impact on cost structure, especially in
uncommon words or phrases, such as expands the potential for commercial large deployments. Sound quality is
slang references, industry terms or to applications where actual recordings are important because it affects customer
modify the default pronunciation of either too costly or simply impractical perception. Each engine deployed today
certain words. The dictionary is also due to the dynamic changes or volume has differences, depending on which
most helpful with certain derivative of changes. TTS is used in applications language is used and the nature of the
languages like French, Canadian English to provide directory assistance, account applications. Let Nortel, with many
and Mexican Spanish. Certain words are status, theater seating availability or even years of speech experience, help you
specific to these regions and may not be driving directions. Many organizations decide the best solution for your
known to the TTS engine. are now using this high-quality method business.
Consequently, they can be added to the to substitute greater proportions of their
user dictionary for proper articulation. studio-recorded prompts. Since the Why Text-to-Speech? You don't have to
Isn't it nice to know you can make an sound is so good, it enables new wonder any more. With amazingly
exception for your business? applications to come to market that natural-sounding speech, created in
much sooner with the recording process software, a whole new horizon of
Create a Custom Voice and cost eliminated. Future changes to automated possibilities has come over
scripts can also be added using the same the horizon. Nortel can let you listen to
Ever listen to the radio or television and voice to further ensure continuity. the difference.
when you hear a certain voice, you Change the text, and the prompts are
immediately identify the organization, ready—it's that simple. Languages
because of their spokesperson's voice?
Why not extend that signature to your The Nortel Difference Nortel's integration of speech engines
telephony applications. TTS can enable within its Media Processing Server
this through the creation of a custom Nortel is both a pioneer in the (MPS) platform is language
voice. With 20 to 40 hours of voice deployment of Text-to-Speech and a independent. Therefore, any language
samples from the person whose voice is leader in platform flexibility. With the speech vendor supports is readily
being captured and several weeks of deployments of all the latest, best-in- available for use in a customer
tuning to create the voice model, the breed TTS engines, each with their own deployment. Nortel has speech
voice can be ready to speak all your unique specialties and differences, recognition deployments in 22 countries
automated prompts, messages, and any Nortel can assist in helping you select using native languages. In some cases,
other dynamic data required. A news the application capability that best fits multiple native languages are deployed
media company might want one of their your needs and support your as part of the same application (e.g.,
network anchors to support a telephone requirements. The speech experts at French and English in Canada) for a
news line. Or a Web merchant with a Nortel will match the right prompt total of 23 languages.
telephone channel could use their voice gender to pre-existing conditions.
television personality to route and It can be most distracting to a caller if
handle sales calls, service requests, and the voice gender changes during a call;
other automated services. Think of the also, if the volume of the speech
possibilities to extend your brand and delivered differs during playback. Some
leverage the quality of service delivered engines may have limited genders
to every customer. available (female vs. male) for certain
languages. Is there a programming
preference of UNIX or Windows? Do
you prefer VXML or other markup
languages? This is important because of
issues relating to density and cost.
Features Available in Many TTS > Job postings, benefits
Engines Deployed by Nortel announcements and other HR-
related messages
> Amazingly high-quality speech right > Theater seating options
out of the box. Nortel platforms > Appointment reminders of date
support a wide array of best-in- and time
breed TTS engines. You'll ask,
“Who is that voice?”
> Multiple voices are available Business Advantages
> Many languages supported in male
or female voices. Choose the one > Additional service offers
that best suits your business. > New capability for automation,
> Nortel Speech Server environment including revenue-enhancing
supports multiple languages and services
Large Vocabulary Recognition > Reduced expenses via a lower cost
(LVR) simultaneously. per call
> Custom voices available for > Increased return on call center
branding and personality matching investment
> Understanding of word context as > Increased customer satisfaction
they appear in a sentence and ability > Better caller experience
to distinguish between words with > Opportunity for unique offerings
identical spelling, but different > Increased competitive
pronunciations differentiation
> Unrestricted vocabulary Benefits of Text-to-Speech
> Adjustable volume, pitch and speed Technology
settings for increased control of
applications > Cuts the costs of voice-driven
> Extensive graphical dictionary and applications by reducing or
development tools eliminating the need for pre-
> VoiceXML, CCXML and SCXML recorded prompts
support > Adds flexibility and speed to
service of automated applications
> Even limited vocabulary routines
Uses like amounts, numbers and dates
can benefit from TTS because it
> Email filter and reader when allows additions to the application
connected to a Unified Messaging at a later date quickly with the same
product or standalone application voice, maintaining your corporate
> News feeds and other dynamic personality
textual information > The ability to seamlessly integrate
> Stock quotes and market indices with pre-recorded prompts
> Business listings, addresses, > A more personalized interaction
locations and directions that can reflect your business
> Order management, status and persona
logistical information > Improved customer satisfaction
> Product information, instructions and acceptance of automation
and trouble-shooting replies > Technologically advanced for a
> Bill pay vendor names, status, competitive advantage
amounts and reject information
In the United States: In Asia Pacific:
Nortel Nortel
35 Davis Drive Nortel Networks Centre
Research Triangle Park, NC 27709 USA 1 Innovation Drive
Macquarie University Research Park
In Canada: Macquarie Park, NSW 2109
Nortel Australia
8200 Dixie Road, Suite 100 Tel +61 2 8870 5000
Brampton, Ontario L6T 5P6 Canada
In Greater China:
In Caribbean and Latin America: Nortel
Nortel Sun Dong An Plaza, 138 Wang Fu Jing
1500 Concorde Terrace Street
Sunrise, FL 33323 USA Beijing 100006, China
Phone: (86) 10 6510 8000
In Europe:
Nortel
Maidenhead Office Park, Westacott Way
Maidenhead Berkshire SL6 3QH UK
Phone: 00800 8008 9009 or
+44 (0) 870-907-9009

Nortel is a recognized leader in delivering communications capabilities that enhance


the human experience, ignite and power global commerce, and secure and protect the
world’s most critical information. Serving both service provider and enterprise cus-
tomers, Nortel delivers innovative technology solutions encompassing end-to-end
broadband, Voice over IP, multimedia services and applications, and wireless broad-
band designed to help people solve the world’s greatest challenges. Nortel does busi-
ness in more than 150 countries. For more information, visit Nortel on the Web at
www.nortel.com.

For more information, contact your Nortel representative, or call 1-800-4 NORTEL
or 1-800-466-7835 from anywhere in North America.

Nortel, the Nortel logo, Nortel Business Made Simple and the Globemark are trade-
marks of Nortel Networks. All other trademarks are the property of their owners.

Copyright © 2007 Nortel Networks. All rights reserved. Information in this docu-
ment is subject to change without notice. Nortel assumes no responsibility for any
errors that may appear in this document.

> BUSINESS MADE SIMPLE

You might also like