A phoneme table defines all the phonemes which are used by a language, together with their properties and the data for their production as sounds.
Generally each language has its own phoneme table, although additional phoneme tables can be used for different voices within the language. These alternatives are referenced from Voice files.
A phoneme table does not need to define all the phonemes used by a language. It can inherit the phonemes from a previously defined phoneme table. For example, a phoneme table may redefine (or add) some of the vowels that it uses, but inherit most of its consonants from a standard set.
The source files for the phoneme data are in the “phsource” directory. “Vowel files”, which are referenced in FMT(), VowelStart(), and VowelEnding() instructions are made using the espeakedit program which is not part of this project. See more in Customization of sound source files.
The phoneme tables are defined in a master phoneme file, named phonemes. This starts with the base phoneme table followed by phoneme tables for other languages and voices. These inherit phonemes from the base table or previously defined tables.
In addition to phoneme definitions, the phoneme file can contain the following:
include <filename>
Includes the text of the specified file at this point. This allows different phoneme tables to be kept in different text files, for convenience. <filename> is a relative path. The included file can itself contain include statements.
phonemetable <name> <parent>
Starts a new phoneme table, and ends the previous table.
<name> Is the name of this phoneme table. This name is used in Voice files.
<parent> Is the name of a previously defined phoneme table whose phoneme definitions are inherited by this one. The name base indicates the first (base) phoneme table. The name _ indicates no parent phoneme table.
A phoneme table contains a list of phoneme definitions. Each starts with the keyword phoneme and the phoneme name (this is the name used in the pronunciation rules in a language's *_rules and *_list files), and ends with the keyword endphoneme.
For example:
phoneme aI
vwl
starttype #a endtype #i
length 230
FMT(vowels/ai)
endphoneme
phoneme s
vls alv sib frc
voicingswitch z
lengthmod 3
Vowelin f1=0 f2=1700 -300 300 f3=-100 80
Vowelout f1=0 f2=1700 -300 250 f3=-100 80 rms=20
IF nextPh(isPause) THEN
WAV(ufric/s_)
ELIF nextPh(p) OR nextPh(t) OR nextPh(k) THEN
WAV(ufric/s!)
ENDIF
WAV(ufric/s)
endphoneme
A phoneme definition contains both static properties and executed instructions. The instructions may contain conditional statements, so that the effect of the phoneme may be different depending on adjacent phonemes, whether the syllable is stressed, etc.
The instructions of a phoneme are interpreted in two different phases. In the first phase, the instructions may change the phoneme and replace it by a different phoneme. In the second phase, instructions are used to produce the sound for the phoneme.
The import_phoneme statement can be used to copy a previously defined phoneme from a specified phoneme table. For example:
phoneme t import_phoneme base/t[ endphoneme
means: phoneme t in this phoneme table is a copy ofphoneme t[ from phoneme table base. A length instruction can be used after import\_phoneme to vary the length from the original.
The phoneme features are described in the Phonemes document. These should be included within the phoneme to fully describe that phoneme.
Within the phoneme definition the following lines may also occur: ((V) indicates only for vowels, (C) only for consonants).
One of these or a manner of articulation from the Phonemes document must be present.
| Type | Description |
|---|---|
liquid | semi-vowels, such as: r, l, j, w |
pause | |
stress | Used for stress symbols, eg: ' , = % |
virtual | Used to represent a class of phonemes. |
| Property | Type | Description |
|---|---|---|
rhotic | (C) | An r type consonant. |
unstressed | (V) | This vowel is always unstressed, unless explicitly marked otherwise. |
nolink | Prevent any linking from the previous phoneme. | |
nopause | Used in a liquid or nasal phoneme to prevent eSpeak NG inserting a short pause if a word starts with this phoneme and the previous word ends with a vowel. | |
trill | (C) | Apply trill to the voicing. |
starttype <phoneme>
Allocates this phoneme to a group so that conditions such as nextPh(#e) can test for any of a group of phonemes. Pre-defined groups for use for vowels are: #@ #a #e #i #o #u. Additional groups can be defined as phonemes with type virtual.
endtype <phoneme>
Allocates this phoneme to a group so that conditions such as prevPh(#e) can test for any of a group of phonemes. Pre-defined groups for use for vowels are: #@ #a #e #i #o #u. Additional groups can be defined as phonemes with type virtual.
lengthmod <integer>
(C) Determines how this consonant affects the length of the previous vowel.
This value is used as index into the length_mods table in the CalcLengths() function in the eSpeak NG program.
voicingswitch <phoneme>
This is used for some languages to change between voiced and unvoiced phonemes.
Phoneme Instructions may be included within conditional statements.
During the first phase of phoneme interpretation, an instruction which causes a change to a different phoneme will terminate the instructions. During the second phase, FMT() and WAV() instructions will terminate the instructions.
length <length>
The relative length of the phoneme in milliseconds. Typical values are about 140 for a short vowel and from 200 to 300 for a long vowel or a diphthong. A length() instruction is needed for vowels. It is optional for consonants.
ipa <ipa string>
In many cases, eSpeak NG makes IPA (International Phonetic Alphabet) phoneme names automatically from eSpeak NG phoneme names. If this is not correct, then the phoneme definition can include an ipa instruction to specify the correct IPA name. IPA strings may include non-ascii characters. They may also include characters specified by their character codes in the form U+ followed by 4 hexadecimal digits. For example a string: aU+0303 indicates ‘a’ with a ‘combining tilde’.
WAV(<wav file>, <amplitude>)
<wav file> is a path to a WAV file (22 kHz, 16 bits, mono) within phsource/ which will be played to produce the sound. This method is used for unvoiced consonants. <wavefile> does not include a .WAV filename extension, although the file to which it refers may or may not have one.
<amplitude> is optional. It is a percentage change to the amplitude of the WAV file. So, WAV(ufric/s, 50) means: play file ‘ufric/s.wav’ at 50% amplitude. Default value is 100.
FMT(<vowel file>, <amplitude>)
<vowel file> is a path to a file (within phsource/) which defines how to generate the sound (a vowel or voiced consonant) from a sequence of formant values. Vowel files are made using the espeakedit program which is not part of this project.
<amplitude> is optional. It is a percentage change to the amplitude of the sound which is synthesized from the FMT() instruction. Default value is 100.
FMT(<vowel file>, <amplitude>) addWav(<wav file>, <amplitude>)
For voiced consonants, a FMT() instruction may be followed by an addWav() instruction. addWav() has the same format as a WAV() instruction, but the WAV file is mixed with the sound which is synthesized from the FMT() instruction.
VowelStart(<vowel file>, <length adjust>)
This is used to modify the start of a vowel when it follows after a sonorant consonant (such as [l] or [j]). It replaces the first frame of the <vowel file> which is specified in a FMT() instruction by this <vowel file>, and adjusts the length of the original by a signed value <length adjust>. The VowelStart() instruction may be specified either in the phoneme definition of the vowel, or in the phoneme definition of the sonorant consonant which precedes the vowel. The former takes precedence.
VowelEnding(<vowel file>, <length adjust>)
This is used to modify the end of a vowel when it is followed by a sonorant consonant (such as [l] or [j]). It is appended to the <vowel file> which is specified in a FMT() instruction by this <vowel file>, and adjusts the length of the original by a signed value <length adjust>. The VowelEnding() instruction may be specified either in the phoneme definition of the vowel, or in the phoneme definition of the sonorant consonant which follows the vowel. The former takes precedence.
Vowelin <vowel transition data>
(C) Specifies the effects of this consonant on the formants of a following vowel. See vowel transitions.
Vowelout <vowel transition data>
(C) Specifies the effects of this consonant on the formants of a preceding vowel. See vowel transitions.
ChangePhoneme(<phoneme>)
Change to the specified phoneme.
ChangeIfDiminished(<phoneme>)
Change to the specified phoneme (such as schwa, @) if this syllable has “diminished” stress.
ChangeIfUnstressed(<phoneme>)
Change to the specified phoneme if this syllable has “diminished” or “unstressed” stress.
ChangeIfNotStressed(<phoneme>)
Change to the specified phoneme if this syllable does not have “primary” stress.
ChangeIfStressed(<phoneme>)
Change to the specified phoneme if this syllable has “primary” stress.
IfNextVowelAppend(<phoneme>)
If the following phoneme is a vowel then this additional phoneme will be inserted before it. Usually it is short pause to distinguish two vowels from diphthongs.
Ends executions of instructions.
CALL <phoneme table>/<phoneme>
Executes the instructions of the specified phoneme.
Phoneme definitions can contain conditional statements such as:
IF <condition> THEN
<statements>
ENDIF
or more generally:
IF <condition> THEN
<statements>
ELIF <condition> THEN
<statements>
...
ELSE
<statements>
ENDIF
where the ELSE and multiple ELSE parts are optional.
Multiple conditions may be joined with AND or OR, but not a mixture of ANDs and ORs.
A condition may be preceded by NOT. For example:
IF <condition> AND NOT <condition> THEN
<statements>
ENDIF
Conditions can be:
thisPh(<attribute>)
Test this current phoneme
prevPh(<attribute>)
Test the previous phoneme
prevPhW(<attribute>)
Test the previous phoneme, but only within the same word. Returns false if there is no previous phoneme in the word.
prev2PhW(<attribute>)
Test the phoneme before the previous phoneme, but only within the same word. Returns false if it is not in this word.
nextPh(<attribute>)
Test the following phoneme
next2Ph(<attribute>)
Test the phoneme after the next phoneme.
nextPhW(<attribute>)
Test the next phoneme, but only within the same word. Returns false if there is no following phoneme in the word.
next2PhW(<attribute>)
Test the phoneme after the next phoneme, but only within the same word. Returns false if not found before the word end.
next3PhW(<attribute>)
Test the third phoneme after the current phoneme, but only within the same word. Returns false if not found before the word end.
nextVowel(<attribute>)
Test the next vowel after the current phoneme, but only within the same word. Returns false if there is none.
prevVowel(<attribute>)
Test the previous vowel before the current phoneme, but only within the same word. Returns false if there is none.
PreVoicing()
This is used as part of the instructions for voiced stop consonants (e.g. [d] and [g]). If true then produce a voiced murmur before the stop.
KlattSynth()
Returns true if the voice is using the Klatt synthesizer rather than the eSpeak synthesizer.
<phoneme name>
True if the phoneme has this phoneme name.
<phoneme group>
True if the phoneme has this starttype (or if it has this endtype if it is used in prevPh()). The pre-defined phoneme groups are #@, #a, #e, #i, #o, #u.
isPause
True if the phoneme is a pause.
isPause2
nextPh(isPause2) is used to test whether the next phoneme is not a vowel or liquid consonant within the same word.
isVowel isNotVowel isLiquid isNasal isVFricative
These test the phoneme type.
isPalatal isRhotic
These test whether the phoneme has this property.
isWordStart notWordStart
These text whether this is the first phoneme in a word.
isWordEnd
True if this is the final phoneme in a word.
isFirstVowel isSecondVowel isFinalVowel
True if this is the First, Second, or Last vowel in a word.
isAfterStress
True if this phoneme is after the stressed vowel in a word.
isVoiced
True if this phoneme is a vowel or a voiced consonant.
isDiminished
True if the syllable stress is “diminished”
isUnstressed
True if the syllable stress is “diminished” or “unstressed”
isNotStressed
True if the syllable stress is not “primary stress”.
isStressed
True if the syllable stress is “primary stress”.
isMaxStress
True if this is the highest stressed syllable in the word.
There are three ways to produce sounds:
WAV() instruction. This is used for unvoiced consonants such as [p], [t] and [s].FMT() instruction. This is used for vowels and also for sonorants such as [l], [j] and [n].WAV file is mixed with a wave generated from formant parameters. Use a FMT() instruction followed by addWav(). This is used for voiced stops and fricatives such as [b], [g], [v] and [z].These specify how a consonant affects an adjacent vowel. A consonant may cause a transition in the vowel's formants as the mouth changes shape between the consonant and the vowel. The following attributes may be specified. Note that the maximum rate of change of formant frequencies is limited by the program.
len=<integer>
Nominal length of the transition in milliseconds. If omitted a default value is used.
rms=<integer>
Adjusts the amplitude of the vowel at the end of the transition. If omitted a default value is used.
f1=<integer>
0: f1 formant frequency unchanged.
1: f1 formant frequency decreases.
2: f1 formant frequency decreases more.
f2=<freq> <min> <max>
<freq>: The frequency towards which the f2 formant moves (Hz).
<min>: Signed integer (Hz). The minimum f2 frequency change.
<max>: Signed integer (Hz). The maximum f2 frequency change.
f3=<change> <amplitude>
<change>: Signed integer (Hz). Frequency change of f3, f4, and f5 formants.
<amplitude>: Amplitude of the f3, f4, and f5 formants at the end of the transition. 100 = no change.
brk
Break. Do not merge the synthesized wave of the consonant into the vowel. This will produce a discontinuity in the formants.
rate
Allow a greater maximum rate of change of formant frequencies.
glstop
Indicates a glottal stop.
On Debian/Ubuntu based systems you can install it with command:
sudo apt-get install espeakedit
If your 64-bit Linux system doesn't have this package, you can download just compiled binary file from here.
To edit *.wav files for sonorant/consonant phonemes you may also need:
On Debian/Ubuntu based systems you can install it with command:
sudo apt-get install audacity
On Debian/Ubuntu based systems you can install it with command:
sudo apt-get install praat
Praat can also provide initial settings for vowel definitions to espeakedit.