E-Speech offers two types of speech output systems: custom-developed speech generation systems and general purpose, full-featured text-to-speech. (Scroll down for demos of the general-purpose TTS systems). We also offer name pronunciation software, and text preprocessing software.

Custom Speech Generation Systems

See our Custom Products page for illustrations of our special-purpose speech generation systems.

Latin-American Spanish Name Synthesis System

Our Latin-American Spanish synthesis system speaks people's names based on E-Speech's proprietary subword concatenation algorithms. The system converts orthography to phonemes and stress marks and then generates speech from its sound inventory.  Here are some examples:

Jose Rodriguez
Juan Rosario Cardenas
Miguel Cruz Gonzalez
Pilar Estella Echevarria
Olivia deJesus Corrales
(listen)

The system can accommodate either text written in conventional Spanish orthography in 8-bit ASCII or text written in 7-bit ASCII without accents.  The system is currently based on a sound inventory that produces excellent or good quality speech for names of almost 90% of people, but it is expandable in inventory size, if equivalent quality is required for more names. It is based on the algorithms described in our Custom Products page.

General Purpose Text-to-Speech Synthesizers

E-Speech offers two general purpose text-to-speech systems for American English. The two systems share the same front end (text normalization and word and name letter-to-phoneme components) but differ in the composition of the sound inventory and the signal processing used to represent the sound inventory. In other words, the two synthesizers have the same front-end accuracy; they just sound different.

•   The original E-Speech synthesizer is based on a sound inventory of English demisyllables and uses LPC coding to represent the demisyllables.

•   The newer E-Speech synthesizer is based on a different speaker's voice, and has an expanded sound inventory, containing context-sensitive demisyllables, and uses RELP coding to represent the demisyllables.

Both E-Speech synthesizers excel at pronouncing names and addresses, at producing speech of high intelligibility,  and at performing the kinds of text normalization needed for many database and information-based applications.  For these reasons, they are ideal for many telephony applications.  Listen to some of the sample applications below, or contact us to find out how E-Speech synthesis can help with your applications.

Samples of the Original E-Speech Synthesizer

Hello

Gerhardt Lachapelle
Daphne Murtagh
PETsMART
Daimler-Chrysler
Bionx Corporation
Cafe de Paris
Tiburon Grille
Cabernet Sauvignon
Azalea Court
Bryn Mawr
Skaneateles
Lake Okeechobee
(listen)

Customer Name and Address

2024561414.....................................................................................
8003484047;Sugarbush|County Travel|501||8|Av||Carlyle|||B
2355551234; boogie|shop II the|123||Main|St||Anytown||B
6165555678; floriano|william j II atty||St John|St||Morristown||R

Lists of People's Names

Because of its superior name pronunciation accuracy and segmental intelligibility, the E-Speech synthesizer is ideal for speaking lists of names. For example, a list of local doctors participating in a health plan:

Neurologists in the Oswego area:
Enrique M. Vasquez
John Athanasios
Ehrhard Breneman
Aaron Narayan
Evelyne Glendenning
Louis Migliaccio
Jocelyn Dlugosz
John Szymanski
Jason Haas
Tracee Frazier
Solomon Metcalfe
Twyla Goeke
Malcolm Deleeuw
Shih Hsin Liu
Lucretia Marlowe Lowry
Cyril Maritain
Althea Papadopoulos
Eunice Furey
Yoshioka Furukawa
(listen)

Driving Directions and Traffic Reports

Because E-Speech software is excellent at pronouncing the names of geographical locations, including street and town names, it is ideally suited to speaking driving directions or traffic reports. For example, in combination with our text preprocessors for traffic newsfeeds, it can provide speech output for traffic reports.

Thursday Traffic Report from SCOUT: Report_time: 16:06 Detail: (10:39) IN UNION COUNTY: ON THE ROUTE 78 LOCAL LANES, THE RIGHT LANES ARE CLOSED BOTH WAYS BETWEEN VAUXHALL ROAD AND THE GARDEN STATE PARKWAY. Report_time: 15:45 Detail: INBOUND G.W.B. MINOR. INBOUND LINCOLN TUNNEL AND HOLLAND TUNNEL MINOR. (listen)

Or it can be used for driving directions:

The nearest restaurant on Massachusetts Avenue in Harvard Square to the Hyatt Regency is the Hong Kong. I'll try to find the best way to get there. If you are on the opposite side of the street from the Hyatt Regency, follow the traffic. Bear left at the fork. After you pass Flagg Street on the right, take the next right onto DeWolf Street. After you cross Mt Auburn Street, make an easy right onto Bow Street. After you pass Arrow Street on the right, make the first hard right onto Massachusetts Avenue. The Hong Kong is about twenty yards down on your right side. That's the end of the directions. (listen)

Stock Market Reports

E-Speech software is excellent for speech recognition or synthesis for stock applications. For example, it pronounces the names of most Nasdaq companies accurately, and because our letter-to-phoneme software is rule-based, it can accurately pronounce names of many new companies. For example:

Athena
Centillium
Cholestech
Easylink
Plumtree
Kyphon
Synaptics
Viacell
Virologic
(listen)

Email

Our email preprocessor converts email format into a text format that is more appropriate for speech synthesis.

Received: from smtp02.mrf.mail.rcn.net (smtp02.mrf.mail.rcn.net [207.172.4.61]) by pluto.njcc.com (8.8.7/8.8.3) with ESMTP id KAA04195 for ; Fri, 1 Mar 2002 10:33:53 -0500 (EST) Received: from 207-172-188-244.s244.tnt2.nywnj.ny.dialup.rcn.com ([207.172.188.244] helo=foobar) by smtp02.mrf.mail.rcn.net with smtp (Exim 3.33 #10) id 16gp5a-0001CB-00 for ewinslow@xyz.com; Fri, 01 Mar 2002 10:36:55 -0500
From: "James Pierson" [jpierson@limbo.njit.edu]
To: "Emily Winslow" [ewinslow@xyz.com]
Subject: Today's meeting
Date: Fri, 1 Mar 2002 10:29:02 -0500
Message-ID: <1GBEBIEGLGEANBNBJDFLCOEGFCAAA.jpierson@limbo.njit.edu>
MIME-Version: 1.0
Hi Emily,
I'm so glad it's Friday :)
I'm confirming our meeting for today at 3:30 in Room 300. See you then.

Jim
James Pierson     jpierson@limbo.njit.edu
Tel: 1 (609) 923-4567
Fax: 1 (609) 923-4568
(listen)

Samples of the E-Speech RELP Synthesizer

Allan Chancer, 456 Main St., Parsippany, NJ.
C. C. Feliciano, 2648 Rockaway Blvd., Piscataway, NJ.
Morningstar Multimedia, Inc., 29 W. Ridgewood Ave., Ridgewood, 07450.
(listen)

Please contact us at einfo@espeech.com for licensing and pricing information.