Using mbrola voices with either espeak or festival
Dave
freebsd01 at dgmm.net
Mon Jun 23 13:39:50 UTC 2014
First off....
FreeBSD testbox.asgard.uk 9.2-RELEASE-p7 FreeBSD 9.2-RELEASE-p7 #0: Tue Jun 3
11:05:13 UTC 2014 root at amd64-
builder.daemonology.net:/usr/obj/usr/src/sys/GENERIC amd64
I'm got both espeak and festival working using their default and supplied
voices but can't get either to use any of the mbrola voices.
mbrola works standalone...
testbox:/usr/local/share/mbrolavox/en1 %mbrola ./en1 ./TEST/mbrola.pho
~/test.wav && mplayer ~/test.wav
...at which point mplayer plays the .wav file
Using espeak...
espeak -v mb-en1 "this is a test" | mbrola -e /usr/local/share/mbrolavox/en1/
- - > ~/fred.raw
Binary number format error
You are probably using a version of /usr/local/share/mbrolavox/en1/
incompatible
with your machine architecture.
Get the right one from the MBROLA project homepage :
http://tcts.fpms.ac.be/synthesis
mbrowrap error: /proc is unaccessible: No such file or directory
mbrola voice not found
espeak speaks using the default voice at about half speed in a very deep bass
voice.
mbrolavox was installed from ports and /proc is mounted
testbox:/home/dave %cat /etc/fstab
# Device Mountpoint FStype Options Dump Pass#
/dev/ada1p2 / ufs rw 1 1
/dev/ada1p3 none swap sw 0 0
/dev/ada1p4 /var ufs rw 2 2
/dev/ada1p5 /tmp ufs rw 2 2
/dev/ada1p6 /usr ufs rw 2 2
proc /proc procfs rw,noauto,0,0
Likewise, using this...
testbox:/home/dave %espeak -v us-mbrola-2 "this is a test"
mbrowrap error: /proc is unaccessible: No such file or directory
mbrola voice not found
...also speaks slow and basso whereas...
espeak -v english "this is a test"
...works fine
Ports installed are...
%pkg info | grep espeak
espeak-1.47.11 Software speech synthesizer
%pkg info | grep fest
festdoc-1.4.0.1 Documentation for Festival
festival-2.1_1 Multi-lingual speech synthesis system
festival-freebsoft-utils-0.10_2 Festival utilities focused on interaction with
Speech Dispatcher
festlex-cmu-1.95 CMU American English pronunciation dictionary
for Festival
festlex-oald-1.4.1_1 Oxford Advanced Learner's pronunciation
dictionary for Festival
festlex-poslex-1.4.1_2 English lexicon for Festival
festvox-don-1.4.0_2 British English male voice for Festival speech
synthesis system
festvox-el11-1.4.0_2 Castilian Spanish male voice for Festival
festvox-kal16-1.4.0 American English male voice, sampled at 16 kHz,
for Festival
festvox-ked16-1.4.0_1 American English male voice, sampled at 16 kHz,
for Festival
festvox-rab16-1.4.1_2 16 kHz British English male voice for Festival
festvox-us1-mbrola-20041012_1 The MBROLA us1 voice (American English, female)
for Festival
festvox-us2-mbrola-20041012_1 The MBROLA us2 voice (American English, low-
pitched male) for Festival
festvox-us3-mbrola-20041012_1 The MBROLA us3 voice (American English, high-
pitched male) for Festival
%pkg info | grep mbrol
festvox-us1-mbrola-20041012_1 The MBROLA us1 voice (American English, female)
for Festival
festvox-us2-mbrola-20041012_1 The MBROLA us2 voice (American English, low-
pitched male) for Festival
festvox-us3-mbrola-20041012_1 The MBROLA us3 voice (American English, high-
pitched male) for Festival
linux-mbrola-3.0.1 MBROLA speech synthesizer for Linux
mbrolavox-3.0.1_1 Voices for MBROLA voice synthesizer
espeak is my preference as I'd like to convert a couple of ebooks for a non-
technical friend who doesn't have a PC so the option to write to a .wav file
and split on the fly in 30 minute chunks is a nice plus but not essential. It
might even be better to split the text files into separate chapters anyway
since there's only a couple of plaintext books to convert. There may be more
later now that I told him about Project Gutenberg :-)
More information about the freebsd-questions
mailing list