Appendix C: Converting Spanish Text

Text characters are used during normal operation of the TSC to specify the message that is to be spoken. The complete set of text characters includes all "printable" ASCII characters, i.e., those having hexadecimal codes between 20 and 7E and relevant ANSI characters.

Text character input is selected using the control sequence <ESC>[0I.

Note: Text character input is the default condition.

Because the standard seven bit ASCII alphabet does not have all characters required for Spanish spelling, incoming text must meet a set of input conventions for the words to be pronounced properly.

These are summarized below.

1. Upper and lower case "ñ" must be input as "~N" and "~n" respectively. For example "niño must be formatted as "ni~no".

2. The acute accent must be represented by placing it before the accented letter. Thus "dí'" must be represented by "d'i". To accent a diphthong, the stress mark must be entered between the first and second vowel of the diphthong such as in di'o.

3. The inverted question mark(¿) and exclamation(¡) point used to introduce questions, commands and exclamations in Spanish should either be omitted or marked by the sentence final question mark ("?") or the sentence final exclamation point ("!"). Thus, "?Est'a en casa?" and "!De veras!" are acceptable, as are "Est'a en casa?" and "De veras!".

4. Dieresis over the letter "u" is not representable. Words like "vergüenza" must be spelled "ver`guenza" without the dieresis.

The text characters "a" through "z" are the most used text characters. They are used for specifying words that are composed entirely of lower case letters.

Example: A word composed entirely of lower case alphabetics.

Spelling: repuesto

Input: repuesto

Upper Case Alphabetic characters

The text characters "A" through "Z" may be used to capitalize words, as at the start of a sentence or a proper name. They may also be used for acronyms, as described below.

Example: A word containing an initial upper case letter.

Spelling: Juana

Input: Juana

Tilde Character

The two character sequence of tilde followed by lower case "n" is interpreted by the TSC as "ñ". Tilde followed by "N" is interpreted as its upper case counterpart.

Spelling: niño

Input: ni~no

Apostrophe Character

The two character sequence of apostrophe followed by any vowel is interpreted by the TSC as the vowel marked with the acute accent.

Spelling: viví'

Input: viv'i

Grave Accent Character

The two character sequence of grave accent followed by lower case "u" is interpreted by the TSC as "ü".

Spelling: vergüenza

Input: verg`uenza

Word Separators

Words are usually separated from each other by one or more characters, but can also be separated by any character that is not "A" through "Z", "a" through "z", apostrophe "'" before a vowel, or tilde "~" before "n" or "N". Multiple consecutive <SP> characters are permissible.

The maximum allowed length of a word is 29 characters. If a word separator has not been encountered by the 29th character of a word, the word is terminated automatically at the 29th letter, and the remainder of the word is treated as a new word.

One Letter Words

The name of a letter is pronounced when a word consisting of a single letter "a" through "x" or "z" is encountered in either upper or lower case. The letters "y" and "Y" are not included since "y" is a valid word which is pronounced "i" rather than as "i griega", the name of the letter.

Example:

Input: b <SP> x <SP> W <SP> I <SP> a <SP> y <SP> f

Speech: be ekis uve doble i a i efe

Acronyms

Acronyms may appear in the input text. An acronym is defined as a word containing at least two letters, all of which are upper case. The TSC speaks each individual letter of an acronym.

Example:

Spelling: Es un producto de SPI.

Input: Es <SP> un <SP> producto <SP> de SPI.

Speech: Es un producto de ese pe i.

Abbreviations

The abbreviations in the following list may appear in the input text. An abbreviation must have at least one lower case letter in it, otherwise it is recognized as an acronym (described above) instead of an abbreviation. The abbreviations of month names in upper case will be expanded in special date formats (see below).

Abbreviations with following periods will be recognized without the period where this follows common Spanish usage. International metric measures will be recognized with or without the period following the abbreviation.

Any abbreviations other than those listed below will not be recognized as abbreviations. They will be spoken as if they are words.

Abreviation	Speech
a.m.	a eme
ago.	agosto
abr.	abril
aptdo.	apartado
admon.	Administración
Av.	Avenida
Blvd.	Boulevard
c.u.	ciudad universitaria
cia.	compañí'a
c/.	calle
cm	centímetros
c/c.	cuenta corriente
crec	creciente
c/u.	cada uno
cta.	cuenta
cap.	capítulo
cte.	corriente
cc	centímetros cúbicos
cts.	céntimos
cf.	confróntese
d.b.	decibelio(s)
dB	decibelio(s)
Dña.	Doña
dcha.	derecha
dom.	domingo
dcho.	Derecho
Dr.	Doctor
dept./dpto.	Departamento
Dra.	Doctora
dic.	diciembre
ef.	efectos
ej.	ejemplo
etc.	etcétera
Excma.	Excelentísima
Excmo.	Excelentísimo
f.c.	ferrocarril
fol.	folio
feb.	febrero
Fr.	Fray
fig.	figura
g	gramo
gr	gramo
Gen.	General
gral.	general
Gob.	Gobernador
hr.	hora(s)
hz	hertzios
hrs.	horas
id.	ídem
it.	ítem
Ilma.	Ilustri’sima
izq.	Izquierdo
Ilmo.	Ilustrísimo
imp.	Imprenta
izqda.	Izquierda
izqdo.	Izquierdo
jr.	Junior
jul.	julio
juev.	Jueves
jun.	Junio
kg	kilogramo(s)
kgr	kilogramo(s)
kHz	kilohertzio
km	kilómetros
km/h	kilómetros por hora
kWh	kilovatiohoras
l	litro(s)
Lcdo.	Licenciado
lb.	Libra
Lic.	Licenciado
lbs.	Libras
lun.	lunes
Lcda.	Licenciada
m	metro
min.	minuto(s)
m/s	metros por segundo
mins.	minutos
may.	mayo
misc.	miscelánea
meng.	menguante
ml	mililitro
mg	miligramos
mm	milímetros
mhz	megahertzio
Mons.	Monseñor
mierc.	miércoles
Mtro.	Maestro
n.b.	nota bene
ntro.	nuestro
nov.	noviembre
núm.	Número
ntra.	Nuestra
oct.	octubre
ozs.	Onzas
onz.	Onza
pág.	Página
Prof.	Profesor
p.d.	posdata
pról.	Prólogo

p.ej.	por ejemplo
Prov.	Provincia
p.m.	pe eme
pta.	Peseta
p.p.	por poder
ptas.	Pesetas
pl.	Plaza
pts.	Pesetas
Rev.	Reverendo
sg.	Segundo
s/n.	sin número
sig.	siguiente
sab.	sábado
sigs.	siguientes
seg.	segundo(s)
Sr.	Señor
segs.	segundos
Sra.	Señora
Sen.	Senador
Sres.	Señores
sep.	septiembre
Srta.	Señorita
sept.	septiembre
Sta.	Santa
tel.	teléfono
tm.	tonelada
ud.	usted
Univ.	Universidad
uds.	ustedes
v.gr.	verbigracia
vol.	volúmen
vds.	ustedes
vols.	volúmenes
vier.	viernes
vs.	versus

Example:

Input: El Dr. García vive en la Av Madrid.

Speech: El Doctor García vive en la Avenida Madrid.

Abbreviations and the Ends of Sentences

The period at the end of most abbreviations acts as a sentence terminator if and only if the next non-space character is an upper case letter. However, for those abbreviations which are usually followed by a person's name, the period never acts as a sentence terminator. These abbreviations are:

Abreviation	Speech
Dr.	Doctor
Mons.	Monseñor
Dra.	Doctora
Mtro.	Maestro
Dña.	Doña
Prof.	Profesor
Excma.	Excelentísima
Rev.	Reverendo
Excmo.	Excelentísimo
Sen.	Senador
Fr.	Fray
Sr.	Señor
Gen.	General
Sra.	Señora
Gob.	Gobernador
Sres.	Señores
Ilma.	Ilustrísima
Srta.	Señorita
Ilmo.	Ilustrísimo
Sta.	Santa

Example:

Input: El Sr. Rodríguez se comío dos kg.. Entonces?

Speech: EL Señor Rodríguez se comío dos kilogramos. Entonces?

Abbreviations Following Numbers

Abbreviations following numbers will agree in number where appropriate. If the number is "1", the following abbreviations will be pronounced in its singular form. All other numbers are followed by the plural pronunciation of the abbreviation.

Example:

Input: 1 km..

Speech: un kilómetro.

Input: 20 km.

Speech: Veinte kilómetros.

Abbreviations Following Millions

If a number represents an exact number of millions, "de" is pronounced after the number and before the expanded form of the abbreviation.

Examples:

Input: 1.000.000 km..

Speech: un millón de kilómetros.

Input: 200.000.000 > pts..

Speech: doscientos millónes de pesetas.

Input: 1.500.000 m..

Speech: un millón quincientos mil metros.

Special Abbreviations and Acronyms

A short list of abbreviations and acronyms are pronounced in their full forms when they consist of capital letters followed by periods.

S. San

D. Don

S.A. Sociedad Anónima

S.L. Sociedad Limitada

F.F.C.C. Ferrocarriles

E.E.U.U Estados Unidos

K.W.H. kilovatios hora

Words without Vowels

Words that do not contain a vowel (a, e, i, o, u, y) and are not abbreviations are spelled out letter by letter.

Example:

Input: Fué un gran grbld.

Speech: Fué un grán ge erre be ele de.

Cardinal Numbers

Cardinal Numbers may appear in the input text, with optional periods to block off groups of three digits. Numbers using periods as three digit group separators are pronounced using the terms "millón", "mil", and the hundreds forms, as appropriate.

Example:

Input: 5

Speech: cinco

Input: 0

Speech: cero

Input: 15

Speech: quince

Input: 017

Speech: cero diecisiete

Input: 98

Speech: noventa y ocho

Input: +10

Speech: más diez.

Input: 1.990

Speech: mil novecientos noventa

Input: 1.017

Speech: mil diecisiete

Input: 567.890.123

Speech: quinientos sesenta y siete millones, ocho cientos noventa

mil, ciento vientitrés.

Large Numbers without Commas

Numbers that are seven or more digits long and do not contain commas are pronounced digit by digit.

Example:

Input: 1990261

Speech: uno nueve nueve cero dos seis uno

Large Numbers with Commas

A comma-blocked number with 12 or more digits is pronounced as a list of three digit numbers rather than one large number.

Example:

Input: 1,981,234,567,890,123

Speech: uno. novecientos ochenta y uno. doscientos treinta y cuatro...

Decimal Numbers

Numbers containing a period followed by one or more digits are treated as decimal numbers. The integer part is pronounced followed by "punto", and the fractional digits. If there are no more than three fractional digits, these are pronounced as the corresponding cardinal number.

Examples:

Input: 0.7

Speech: cero punto siete

Input: .76034

Speech: punto setenta y seis mil treinta y cuatro

Input: El tiene 10, y Elena tiene 20.

Speech: El tiene diez, y Elena tiene veinte.

Numbers With Ordinalizers

Numbers between "1" and "19" are pronounced in their ordinal forms when followed by "o" for masculine forms and "a" for feminine forms. "1er." and "3er." are pronounced "primero" and "tercero". The superscripted "o", "a" and "r" of standard Spanish orthography cannot be represented in 7-bit ASCII. Numbers greater than 19 are pronounced in their cardinal forms.

Examples:

Input: 3er piso

Speech: tercer piso

Input: 2o. piso.

Speech: segundo piso.

Numbers Followed by Abbreviations for Quantities

Numbers followed by abbreviations for units of measure and monetary amounts will be assigned the appropriate gender. These abbreviations will be recognized with or without a period.

Masculine quantity measures

Unit	Pronunciation
cm	centímetro
cts.	céntimos
dB	decibelio
g	gramo
gr	gramo
Hz	hertzio
kg	kilogramo
kgr	kilogramo(s)
kHz	kilohertzio
km	kilómetro
l	litro
m	metro
mg	miligramo
MHz	megahertzio
min.	minuto(s)
mins.	minutos
ml	milílitro
mm	milímetro
s	segundo
segs.	segundos
sg.	segundo(s)

Feminine quantity measures

Unit	Pronunciation
hr.	hora(s)
hrs.	horas
lb.	libra
lbs.	libras
onz.	onza(s)
ozs.	onzas
pág.	página
pta.	peseta
ptas.	pesetas
pts.	pesetas
tm.	tonelada

Examples:

Input: 21 km..

Speech: veintiún kilo ’metros

Input: 200.000 pts.

Speech: dos cientas mil pesetas.

Time of Day

Digit pairs separated by a colon are interpreted as designating time of day when they appear in the format "hh:mm" or "hh:mm:ss" where "h","m", and "s" are digits representing hours, minutes and seconds. The hours digits are pronounced in their feminine form. the minutes and second digits are pronounced in their masculine forms. The sequence "nn:00" is converted to "nn en punto", as a special case. If other than exactly two digits follow it, the colon is pronounced as "dos puntos."

Examples:

Input: 12:35

Speech: doce treinta y cinco

Input: 3:00

Speech: tres en punto

Input: 3:05 a.m..

Speech: tres cero cinco a eme

Input: 3:00 a.m..

Speech: tres en punto a eme.

Input: 23:01 pm..

Speech: veinte tres cero cinco pe eme

Input: 1:25:20 a.m..

Speech: una veinte cinco veinte a eme

Input: 1:2:204:25

Speech: uno, dos puntos, dos, dos puntos, dos cero cuatro,

veinticuatro

Dates

Dates are recognized when they consist of one or two digits representing the day of the month, a hyphen, a three letter abbreviation for the month name, another hyphen, and by two or four digits representing the year. The date is pronounced by saying the number for the day, "de", the month name, "de", followed by a four digit year or "del" if followed by a two digit year. In this format, the abbreviations listed below are recognized for the month names in either upper or lower case.

Time	Unit
abr	abril
ago	agosto
dic	diciembre
ene	enero
feb	febrero
jul	julio
jun	junio
mar	marzo
may	mayo
nov	noviembre
oct	octubre
sep	septiembre

Examples:

Input: 3-sep-84

Speech: tres de septiembre del ochenta y cuatro

Input: 31-dec-1985

Speech: treinta y uno de deciembre de mil novecientos ochenta y

cinco

Dollar Amounts

When a number is preceded by a "$", it is said in its masculine form followed by either "dólar" or "dólares". The number may optionally be followed by the word "millón" or "millones" in which case the phrase is pronounced "millón de dólares" or "millones de dólares". If the number has a comma, the first part is pronounced followed by the word "dólares" and the second part is pronounced followed by the word "centavos". In the following examples, the Digit Pronunciation Rule is OFF and the Full Number Pronunciation Rule is ON.

Examples:

Input: $1

Speech: un dólar

Input: $12

Speech: doce dólares

Input: $135

Speech: ciento treinta y cinco dólares

Input: $1,10

Speech: un dólar y diez centavos

Input: $0,10

Speech: cero dólares y diez centavos

Input: $1 million

Speech: un millón de dólares

Input: $1.256

Speech: mil dos cientos cincuenta y seis dólares

Input: $

Speech: dollar

Garbage Input

Although the input text is usually human-readable, it doesn't have to be. Text that is totally unreadable can be sent to the TSC, and it will make an attempt to say it.

Terminal Punctuation

The period, question mark, and exclamation point text characters are used as sentence terminators. These characters cause a short pause to occur in the speech output at the end of the sentence, and they cause appropriate pitch intonation to occur. If a period is immediately followed by an alphabetic character, the period is ignored. If a single letter is followed by a period

followed by fewer than two <SP> characters, the period is ignored. This is useful for forms such as "la organización de la O.N.U." . Real end-of-sentence periods should be followed by two <SP> characters to ensure that they are interpreted as periods. (Only a single <SP> is shown in most examples here.)

Sentence Internal Punctuation

The text characters listed below can be used freely throughout a sentence. They cause a short pause and a "comma-type" pitch intonation to occur in the speech.

Character	ASCII name
,	Comma
;	Semicolon
(	Opening Parenthesis
[	Opening Bracket
{	Opening Brace

The TEXT CHARACTERS listed below likewise can be used freely throughout a sentence. They cause a short pause and a "comma-type" pitch intonation to occur in the speech only if the next non-space character following them is alphabetic or a digit.

Character	ASCII Name
"	Quotation Mark
)	Closing Parenthesis
]	Closing Bracket
}	Closing Brace

Example:

Input: Enrique (el abogadodeJuan) ten'ia influencia con el juez.

Speech: Enrique, el abogado de Juan, tenía influencia con

el juez.

Hyphen Characters

The hyphen text character is usually treated in the same way as a <SP>, acting as a word separator but otherwise ignored. However, if the next non-space character is a digit character ("0" through "9"), the hyphen is handled differently. See the description of the Minus Sign Switch below for details. Hyphens are additionally used to separate days, months, and years in the date format

Miscellaneous Punctuation Characters

The punctuation characters listed below can be used freely throughout the input text. When encountered by the TSC, they are pronounced as indicated.

Character	ASCII Name	Pronunciation
#	Number Sign	número
%	Percent Sign	por ciento
&	Ampersand	y
+	Plus	más
<	Less Than	menor que
=	Equals	igual
¬	Greater Than	mayor que
@	Commercial At	arroba
/	Slash	barra
\	Reverse Slash	barra invertida
^	Circumflex	acento circunflejo
_	Underline	subrayado
|	Vertical Line	barra vertical

Example:

Input: Juan trabaja en la compañía de Aceros & Metales.

Speech: Juan trabaja en la compañía de Aceros y Metales.

Input: Hoy es 13/4/85.

Speech: Hoy es trece barra cuatro barra ochenta y cinco.

Phoneme characters are used when it is desired to specify the precise sounds that are to be spoken, or when it is desired to provide syntactic information to improve speech quality. They might be used, for example, to pronounce correctly a word that would otherwise be phonemicized incorrectly by the TSC when entered as text characters. Phoneme character input is selected using the control sequence "<ESC>[1I". To return to text character input use the control sequence "<ESC>[0I".

There are two types of phonemes, segmental phonemes and non-segmental phonemes. Segmental phonemes are defined as the characters used to represent the sound units of speech. They correspond to the units denoted by the terms "phoneme", "allophone", and "diphthong" in normal linguistic usage. The <SP> character is included in this group in order to permit the specification of silence.

The table on the proceeding page shows the segmental phoneme alphabet, listing the ASCII character that represents the sound produced in speaking the underlined letters in the word to the right.

Allophonic phonemes are usually not needed in the phoneme character input mode because they are generated automatically by allophonic rules in the TSC software. However, they can be useful in special cases, an example being the imitation of non-Spanish words (e.g. the pronounciation of "Jacque" can be approximated by "<ESC>[1IyAK<ESC>[0I").

The allophonic phonemes are listed below.

• "b" is sometimes generated from "B".

• "d" is sometimes generated from "D".

• "g" is sometimes generated from "G".

• "y" is sometimes generated from "Y".

• "~" is sometimes generated from "N".

• "a" is sometimes generated from "AI".

• "e" is sometimes generated from "EI".

• "h" is sometimes generated from "IA".

• "i" is sometimes generated from "IE".

• "j" is sometimes generated from "IO".

• "k" is sometimes generated from "IU".

• "m" is sometimes generated from "UA".

• "o" is sometimes generated from "OI".

• "p" is sometimes generated from "UE".

• "q" is sometimes generated from "UI".

• "t" is sometimes generated from "UO".

• "u" is sometimes generated from "OU".

• "v" is sometimes generated from "AU".

• "w" is sometimes generated from "EU".

Non-segmental phonemes do not represent a particular sound (as characterized by a certain frequency spectrum) but they do have an effect on the sounds produced by adjacent segmental phonemes. They are described below. When phoneme character input is selected, the TSC discards any data character it encounters that is not a valid segmental or non-segmental phoneme.

Segmental Input

Each segmental phoneme is assigned a pitch and duration value by the TSC. It is possible to override this assignment by specifying pitch and duration of each phoneme as it is sent to the TSC. This is done by following the phoneme with a "/", the pitch and duration separated by a ";", and then a terminating "/". The pitch is specified in Hertz. The duration is specified as an integer representing a multiple of 10 milliseconds for the duration.

Either the pitch or the duration field may be left empty, in which case that value will be assigned automatically by the TSC. The following limits must be observed for pitch and duration. If either value is specified outside of these ranges, the pitch and duration will both be assigned automatically by the TSC.

Pitch: Minimum: 50 (or 0) Maximum: 200

Duration: Minimum: 1 Maximum: 60

A pitch value of zero means to use a random noise source instead of a periodic impulse train as the voicing source. The resulting speech sounds like whispering.

Example:

Input: Su nombre es <ESC>[1I &L/;10/A/85;13/N/80/A <ESC>[0I.

In this example, phoneme "L" has been given a duration of 10 * 10 or 100 milliseconds, and its pitch is assigned by the TSC. Phoneme "A" has a pitch of 85 Hertz and its duration is 0.13 seconds. "N" has a pitch of 80 Hertz and its duration is assigned by the TSC.

In phoneme input mode, the F0 and duration values for segmental phonemes may also be specified as deviations from the default values. Incremental segmental input is bounded by the "!" character rather than to "/" character. The incremental values are all entered with an offset of 100 decimal, so that both increments and decrements are positive. For example, to increase pitch by 10 Hertz:

&L!110;!...

and to decrease pitch by 10 Hertz:

&L!90;!...

Vowels and Diphthongs

Each non-segmental phoneme is described in detail below. The first line of each description gives the ASCII character that is used to represent that phoneme. Examples are shown to further clarify the descriptions. The symbol "->" is used to mean "is translated into the following phonemes."

Stress

ASCII: '

This phoneme is used to mark a syllable of a word that is to receive

word-level stress. It must immediately precede the vowel phoneme in

the syllable to be stressed.

Example: tenía -> &TEN'IA

Word Boundary

ASCII: &

This phoneme is used to mark the beginning of a word. It should appear immediately before the first segmental phoneme of the word. There are no required inter-word phonemes of any kind, word boundary or syntactic. For instance, "Vino al pueblo." could be entered in phoneme mode either as "B'INOALP'pbLO." or as "&B'INO&AL&P'pbLO." However, in certain cases a word boundary phoneme may be required in order for correct speech to be produced. Whenever it is unclear whether or not a word boundary phoneme should be included, it is best to include it. If there are other non-segmental phonemes preceding a word, they must precede the word boundary phoneme. When operating with word prosody, the TSC inserts a "," PHONEME following each word. For this reason, word boundary phonemes should be included in phoneme character input if word prosody is to be used. Otherwise, the TSC will arbitrarily break words at a maximum word size of about 29 characters in order to insert the ",".

Example: tenía -> &TEN'IA

Period

ASCII: .

This phoneme marks the end of a declarative sentence. It causes an end-of-sentence intonation gesture in the speech pitch pattern, and is followed by a pause. A longer pause, such as at the end of a paragraph, can be accomplished by adding additional periods.

Example: Vino a casa. -> &B'INO&A&K'ASA.

Question Mark

ASCII: ?

This phoneme should be used to mark the end of a question which is in statement form (e.g. "Vino a casa?"). It is optional (and not recommended) for questions beginning with "quién", "qué", "cuándo", "dónde", "por que ’", "cómo", "cuál" and other interrogative pronouns (e.g. "Dónde está’?"). It causes an end-of-question-sentence intonation gesture in the pitch pattern followed by a pause.

Examples: Vino a casa? -> &B'INO&A&K'ASA?

Do’nde esta’? -> &D’ONDE&EST’A.

or &D’ONDE&EST’A?

Esta ’ en casa? -> &EST’A&EN&K’ASA?

Comma

ASCII: ,

This phoneme should be placed wherever an orthographic comma, dash, semicolon, or colon would be placed. It should be used at the end of clauses, long phrases (where appropriate), between items in a list, and to break up any large unit into smaller sensible chunks for ease of comprehension. It causes a short pause in the speech. A longer pause can be accomplished by using two or more consecutive commas.

Example: Sí,ésto es. -> &S'I, 'ESTO&'ES.

Silence

ASCII: <SP>

This phoneme is classed as both segmental and non-segmental. It causes a short silence to be produced in the speech. The silence duration varies but is typically about 120 ms. It may be used to emphasize other punctuation marks, such as making a period sound more like a paragraph break.

Note: Do not confuse the <SP> phoneme with the <SP> text character. The former causes a short pause while the latter only serves as a word separator.

Example:

In the phonemicization of "Después de hablar con Andrés, Juan y María, Jaime regresó a casa.", an <SP> phoneme might be placed after the last comma. An alternative would be to change the last comma in the phonemic representation to a double comma ",,".