Closed mirfan899 closed 4 years ago
When I try to run make install
following errors occurred.
install: cannot stat './ekho-data/zh.dict': No such file or directory
install: cannot stat './ekho-data/pinyin.index': No such file or directory
install: cannot stat './ekho-data/pinyin.voice': No such file or directory
install: cannot stat './ekho-data/jyutping.index': No such file or directory
install: cannot stat './ekho-data/jyutping.voice': No such file or directory
Makefile:1400: recipe for target 'install-data-hook' failed
make[2]: *** [install-data-hook] Error 1
make[2]: Leaving directory '/home/virtuoso_irfan/ekho'
Makefile:1323: recipe for target 'install-data-am' failed
make[1]: *** [install-data-am] Error 2
make[1]: Leaving directory '/home/virtuoso_irfan/ekho'
Makefile:1268: recipe for target 'install-am' failed
make: *** [install-am] Error 2
put Cantonese voice data to ekho/src/ekho-data/jyutping
before running make install
, run ./ekho 123
and ./ekho -v Cantonese 123
to generate those files. You can also copy those files from a release package.
Now, I'm getting this error when I try to run ./ekho 123
root@instance-2:/home/virtuoso_irfan/ekho# ./ekho 123
Segmentation fault (core dumped)
I've removed the files and run the command again and it works but got this issue.
pa_simple_new() failed: Connection refused
pa_sample_spec(format=3,rate=44100,channels=ch=1)
Fail to init audio stream.
Fail to init sound.
Is pulseaudio service running in your system?
$ ps -ef|grep pulse
hgneng 1432 1 0 14:33 ? 00:00:01 /usr/bin/pulseaudio --start --log-target=syslog
output
root@instance-2:/home/virtuoso_irfan/ekho# ps -ef|grep pulse
root 16221 1 0 07:03 ? 00:00:00 pulseaudio -D
root 16242 29346 0 07:04 pts/0 00:00:00 grep --color=auto pulse
I think it is a system issue. I'm running Ubuntu 18.04 on GCP
When I tried to run pusleaudio --start
It throws the following error
E: [pulseaudio] main.c: --start not supported for system instances.
If you are not running on a desktop, you should not hear any sound. Try to save it to a file, download to local and play:
./ekho -o demo.wav 123
Great, it works. Thanks for quick response.
I've some Cantonese files audio and text files, should I replace jyutping
folder to build custom voice?
Yes, you can. Remember to remove ekho-data/jyutping.* and re-generate them after replacing the voice files.
It seems voice is built using audio files only. How are you doing? Is there a way to use ekho
model in Merlin?
Sorry, there is no way to use ekho model in Merlin.
Is there a way to get the text files of Cantonese wav
files?
http://eguidedog.net/files/jyutping_phon_list.txt (page is encoded in utf-8)
Thanks!! I check the wav
files and they have short duration mostly! I've wav
files with 5-10 seconds long audios. Will ekho work with these files?
No, you should cut the silence in advanced. I suppose each file should be less than 1 second.
I'm trying to use it for Cantonese(https://sourceforge.net/projects/e-guidedog/files/Ekho-Voice-Data/0.2/jyutping-wong-44100-v9.tar.bz2) but did not find where to place the data. Also when I compile it on Ubuntu 18.04 it shows something like this.
Is it installed without errors? Because I don't see any confirmation of ekho is being installed completely.