Skip to content

Latest commit

 

History

History
368 lines (273 loc) · 15.9 KB

mbrola.md

File metadata and controls

368 lines (273 loc) · 15.9 KB
Raw

MBROLA Voices

The MBROLA is an open source speech engine with collection of diphone voices for speech synthesis. Voices are cost-free for non-commerical purposes, but are not open source.

MBROLA does not include any text-to-phoneme translation, so this must be done by another program. eSpeak NG can be used as a front-end to MBROLA, to provide spelling-to-phoneme translation and intonation, which MBROLA then uses to generate speech sound.

Voice Names

To use a MBROLA voice, eSpeak NG needs information to translate from its own phonemes to the equivalent MBROLA phonemes.

The eSpeak NG voices which use MBROLA are named as:

mb-xxN

where xxN is the name of a MBROLA voice (e.g. mb-en1 for the MBROLA en1 English voice). These voice files are in eSpeak NG's folder .../espeak-ng-data/voices/mb (where ... is usually /usr/share, but could be somewhere else).

There are some additional eSpeak NG MBROLA voices, which speak English text using a MBROLA voice for a different language. These contain the name of the MBROLA voice with a suffix -en. For example, the voice mb-de4-en will speak English text with a German accent by using the MBROLA de4 voice.

The following MBROLA voices are available in eSpeak NG1.

MBROLA Voice Language Gender eSpeak Debian Package
af1 Afrikaans male mb-af1 mbrola-af1
ar1 Arabic male mb-ar1
ar2 Arabic male mb-ar2
br1 Brazillian Portugues male mb-br1 mbrola-br1
br2 Brazillian Portugues male mb-br2 mbrola-br2
br3 Brazillian Portugues male mb-br3 mbrola-br3
br4 Brazillian Portugues female mb-br4 mbrola-br4
bz1 Breton female
ca1 Canadian French male mb-ca1
ca2 Canadian French male mb-ca2
cn1 Mandarin Chinese female mb-cn1
cr1 Croatian male mb-cr1 mbrola-cr1
cz1 Czech female mb-cz1
cz2 Czech male mb-cz2 mbrola-cz2
de1 German female mb-de1 mbrola-de1
de2 German male mb-de2 mbrola-de2
de3 German female mb-de3 mbrola-de3
de4 German male mb-de4 mbrola-de4
de5 German female mb-de5 mbrola-de5
de6 German male mb-de6 mbrola-de6
de7 German female mb-de7 mbrola-de7
de8 German-Bavarian male mb-de8
ee1 Estonian male mb-ee1 mbrola-ee1
en1 British English female mb-en1 mbrola-en1
es1 Spanish male mb-es1 mbrola-es1
es2 Spanish male mb-es2 mbrola-es2
es3 Spanish female mb-es3
es4 Spanish male mb-es4
fr1 French male mb-fr1 mbrola-fr1
fr2 French female mb-fr2
fr3 French male mb-fr3
fr4 French female mb-fr4 mbrola-fr4
fr5 Belgian French male mb-fr5
fr6 French male mb-fr6
fr7 Belgian French male mb-fr7
gr1 Greek male mb-gr1 mbrola-gr1
gr2 Greek male mb-gr2 mbrola-gr2
hb1 Hebrew male mb-hb1
hb2 Hebrew female mb-hb2
hn1 Korean male
hu1 Hungarian female mb-hu1 mbrola-hu1
ic1 Icelandic male mb-ic1 mbrola-ic1
in1 Hindi male mb-in1
in2 Hindi female mb-in2
id1 Indonesian male mb-id1 mbrola-id1
ir1 Iranian Persian male mb-ir1 mbrola-ir1
it1 Italian male mb-it1
it2 Italian female mb-it2
it3 Italian male mb-it3 mbrola-it3
it4 Italian female mb-it4 mbrola-it4
jp1 Japanese male mb-jp1
jp2 Japanese female mb-jp2
jp3 Japanese female mb-jp3
la1 Classical Latin male mb-la1 mbrola-la1
lt1 Lithuanian male mb-lt1 mbrola-lt1
lt2 Lithuanian male mb-lt2 mbrola-lt2
ma1 Malay female mb-ma1
mx1 Mexican Spanish male mb-mx1 mbrola-mx1
mx2 Mexican Spanish male mb-mx2 mbrola-mx2
nl1 Dutch male mb-nl1 2
nl2 Dutch male mb-nl2 mbrola-nl2
nl3 Dutch female mb-nl3
nz1 Maori male mb-nz1
pl1 Polish female mb-pl1 mbrola-pl1
pt1 Portuguese female mb-pt1 mbrola-pt1
ro1 Romanian male mb-ro1 mbrola-ro1
sw1 Swedish male mb-sw1 mbrola-sw1
sw2 Swedish female mb-sw2 mbrola-sw2
tl1 Telugu female mb-tl1
tr1 Turkish male mb-tr1 mbrola-tr1
tr2 Turkish female mb-tr2 mbrola-tr2
us1 American English female mb-us1 mbrola-us1
us2 American English male mb-us2 mbrola-us2
us3 American English male mb-us3 mbrola-us3
vz1 Venezuelan Spanish male mb-vz1 mbrola-vz1

eSpeak NG will look for mbrola voices firstly in espeak-ng-data/mbrola and then in /usr/share/mbrola.

The installation instructions below use the MBROLA voice en1 as an example. You can use other mbrola voices for which there is an equivalent eSpeak NG voice in espeak-ng-data/voices/mb.

Windows Installation

The SAPI5 version of eSpeak NG uses the mbrola.dll.

  1. Install eSpeak NG. Include the voice mb-en1 in the list of voices during the eSpeak NG installation.

  2. Install the PC/Windows version of MBROLA (MbrolaTools35.exe) from: http://www.tcts.fpms.ac.be/synthesis/mbrola/bin/pcwin/MbrolaTools35.exe.

  3. Download the en1 or other voice from: https://github.com/numediart/MBROLA-voices

  4. Save downloaded en1 data file into C:/Program Files/eSpeak/espeak-ng-data/mbrola.

  5. Use the voice espeak-MB-EN1 from the list of SAPI5 voices.

Linux Installation

Installation of standard packages

There are standard packages prepared for MBROLA binary and voices on different Linux distributions. On Debian/Ubuntu-like Linux, you can install mbrola using apt-get package manager:

sudo apt-get install mbrola mbrola-en1

where:

  • mbrola is package containing MBROLA executable,
  • mbrola-en1 is mbrola data files for en1 MBROLA voice.

You can check other available voices searching with command:

apt-cache search mbrola

If some MBROLA voice is not available as Debian package, you can download data files from https://github.com/numediart/MBROLA-voices and save them in /usr/share/mbrola/xxN/xxN file, where xxN is code and number of language file.

Installation of MBROLA package from source

  1. Install necessary packages:
sudo apt-get install git make gcc
  1. Clone MBROLA git repository:
git clone https://github.com/numediart/MBROLA.git
cd MBROLA
  1. Compile MBROLA:
make
  1. Install MBROLA:
sudo cp Bin/mbrola /usr/bin/mbrola

  1. Download the en1 or other voice from https://github.com/numediart/MBROLA-voices

  2. Copy the en1 data file to /usr/share/mbrola/en1 folder.

Usage

If you use the eSpeak NG voice such as mb-en1 then eSpeak NG will use the mbrola "en1" voice, e.g.:

espeak-ng -v mb-en1 "Hello world"

To generate mbrola phoneme data (.pho file) you can use:

espeak-ng -v mb-en1 -q --pho "Hello world"

or

espeak-ng -v mb-en1 -q --pho --phonout=out.pho "Hello world"

If you have issues with sound system integration, you can pass MBROLA's output to standard output, which then can be played by ALSA (aplay) or PulseAudio (paplay) player throug the pipe:

    espeak-ng -vmb-en1 --stdout "Hello world"|aplay

Adding new MBROLA voice entry to eSpeak NG

To add new MBROLA voice entry for eSpeak NG you have to:

  1. add MBROLA voice definition file,
  2. add MBROLA phoneme translation file,
  3. update Makefile.am file and compile voice.

These steps are described in details in following sections.

1. Add MBROLA voice definition file

eSpeak NG's voice files for MBROLA voices are in espeak-ng-data/voices/mb folder. Voice definition file is in form mb-xxN and have to contain at least this line: mbrola <voice> <translation>, e.g.

mbrola en1 en1_phtrans

Where:

  • en1 is the name of the MBROLA voice.
  • en1_phtrans is a translation file to convert between eSpeak NG phonemes and the equivalent MBROLA phonemes.

If language code differs between MBROLA voice and eSpeak NG language, then additional line is necessary:

    language xx

where xx is code of the language in eSpeak NG.

Binary xxN_phtrans files are kept in espeak-ng-data/mbrola_ph folder and are generated from phsource/mb/xxN text files, during MBROLA voice compilation.

Additionally MBROLA voice definition file can have other optional parameters, similar to eSpeak NG voices, which are described Voices file.

2. Add MBROLA phoneme translation file

MBROLA phoneme translation files specify translations from eSpeak NG phoneme names to mbrola phoneme names.

The source phoneme translation files are in .../phsource/mbrola folder and their name is in form xxN of referenced MBROLA voice.

Note:

  • eSpeak NG phonemes are referenced from voice files in phsource folder of particular language e.g. ph_english and/or general phonemes file.

  • MBROLA phonemes are usualy listed in README.txt file of MBROLA voice.

Each line in the mbrola phoneme translation file contains:

<control> <espeak ph1> <espeak ph2> <percent> <mbrola ph1> [<mbrola ph2>]

  • <control>
    bit 0 (+1) skip the next phoneme
    bit 1 (+2) match this and previous phoneme
    bit 2 (+4) only at the start of a word
    bit 3 (+8) don't match two phonemes across a word boundary
    bit 4 (+16) add this phoneme name as a prefix to the next phoneme name (used for de4 phoneme prefix '?')
    bit 5 (+32) only in stressed syllable
    bit 6 (+64) only at the end of a word

  • <espeak ph1>
    The eSpeak NG phoneme which is to be translated to an mbrola phoneme.

  • <espeak ph2>
    If this field is not NULL, then the match only occurs if this field matches the next phoneme. If control bit 1 is set, then the previous rather than the next phoneme is matched. This field may also have the following values: VWL
    matches any Vowel phoneme.

  • <percent>
    If this field is zero then only one mbrola phoneme is used. If this field is non-zero, then two mbrola phonemes are used, and this value gives the percentage length of the first mbrola phoneme.

  • <mbrola ph1>
    The mbrola phoneme to which the eSpeak NG phoneme is translated. This field may be NULL.

  • <mbrola ph2>
    The second mbrola phoneme. This field is only used if the <percent> field is not zero.

The list is searched from start to finish, until a match is found. Therefore, a line with more specific match condition should appear before a line which matches the same eSpeak NG phoneme but with a more general condition.

Note:

You can get list (and descriptions) of defined phonemes for particular eSpeak NG language by entering command in phsource folder:

egrep "^phoneme " phonemes ph_english|cut -d$' ' -f2-|sort

where ph_english is phoneme definition for particular language

Note that ph_language file can extend or override phoneme definitions in phonemes file. Translations for all defined phonemes should be given in the mbrola phoneme translation file.

If the language's *_list or *_rules files includes rules to speak words "as English" the mbrola phoneme translation file should include rules which translate English phonemes into near equivalents, so that they can spoken by the MBROLA voice.

For latest Linux verson of mbrola you can get list of diphones for particular language by entering command:

  mbrola -d /usr/share/mbrola/en1/en1

where /usr/share/mbrola/en1/en1 is location of the voice file.

When phoneme translation source file is compiled (look at next section) espeak-ng-data/mbrola_ph/xxN_phtrans file is created.

3. Compile voice and update Makefile.am file

Separate MBROLA voice can be compiled using comand:

espeak-ng --compile-mbrola=<xxN>

where xxN is MBROLA voice name.

Makefile.am is build configuration file which should be extended, to include automatic compilation of newly added MBROLA voice for eSpeak NG.

Search for mbrola: \ line in Makefile.am and add additional line for newly created MBROLA voice, e.g.:

mbrola: \
...
espeak-ng-data/mbrola_ph/xxN_phtrans \

Note that it could be that several voices share the same translation file. Then translation file is named just xx.

When Makefile.am is update, when espek-ng will be compiled, e.g. by entering command automake; make -B newly added MBROLA voice will be compiled automatically.

1: You can get list of configured entries for MBROLA synthesizer with following command in espeak-ng project folder:

cd espeak-ng-data/voices/mb/; for i in $(ls); do printf "$i "; grep name $i|awk '{print $2}'; done

2: This voice has very limited set of diphones and is usable only for reading numbers.