julius-speech / segmentation-kit Goto Github PK
View Code? Open in Web Editor NEWSpeech Segmentation Toolkit using Julius
License: MIT License
Speech Segmentation Toolkit using Julius
License: MIT License
Hi,
How should the data for the mfcc file input be prepared? With librosa? Special parameters to take care of?
Would be very nice to have short guide on how to do this.
Thank you very much.
Hello, I am using julius-speech to do forced alignment.
https://github.com/timmahrt/pyJuliusAlign
My code uses an old perl script segment_julius4.pl
but I would like to use the most recent version.
The most recent one says it takes a command line argument to specify the install location--which my script needs--but it does not. Can we make the following change?
Line 33 of segment_julius.pl
:
## data directory
$datadir = "./wav";
if (defined $ARGV[0]) {
$datadir = $ARGV[0];
}
What do you make of the fact that ty and zy phonemes are not present in hmmdefs_monof_mix16_gid? (from https://julius.osdn.jp/archive/japanese-models.tar.gz I assume)
Finding these phonemes in the segmenter gives the julius error:
Error: voca_load_htkdict: line 2: phone "ty" not found
As a workaround I use dy and sy, respectively.
Log displays strange verbose file path when there are multile wav files.
This issue is not critical, because it is just only annoying.
Here is my directory tree which contains two wav file named a01.wav and a02.wav.
.
├── bin
│ ├── julius-4.3.1
│ ├── julius-4.3.1.exe
│ └── yomi2voca.pl
├── License.md
├── models
│ ├── hmmdefs_monof_mix16_gid.binhmm
│ ├── hmmdefs_ptm_gid.binhmm
│ └── logicalTri
├── README.md
├── segment_julius.pl
└── wav
├── a01.txt
├── a01.wav
├── a02.txt
└── a02.wav
And when I run segment_julius.pl, it outputs like following.
-> perl segment_julius.pl [~/Desktop/sk]
./wav/a01.wav
enter filename->..............................................................................................................................................................................................................................................................................................................................................................................................................................................................enter filename->1 files processed
Result saved in "./wav/a01.lab".
./wav//a02.wav
enter filename->.................................................................................................................................................................................................................................................................................................................enter filename->1 files processed
Result saved in "./wav//a02.lab".
The output contains file path ./wav//a02.wav
, which is verbose.
./wav//a02.wav
is equals to the simplified form ./wav/a02.wav
, and it should output simplified version.
I confirmed this issue in Linux, but I think we can see it also in Windows.
segment_julius.pl
can convert う゛
to b u
, but cannot convert ゔ
.
I want to fix this problem, so could you teach me how to fix.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.