Segmentation Fault for unified parser

[Chaitanya Jadhav]

Hello Team,

I am trying to use unified parser but I am getting a Segmentation fault (core dumped) error.

Following is the output for various arguments:

chaitanya@hp-notebook:~/Downloads/unified$ ./unified-parser नमस्ते 1 0 0 0
Segmentation fault (core dumped)
chaitanya@hp-notebook:~/Downloads/unified$ ./unified-parser नमस्ते 1 1 0 0
Segmentation fault (core dumped)
chaitanya@hp-notebook:~/Downloads/unified$ ./unified-parser नमस्ते 1 2 0 0
Segmentation fault (core dumped)
chaitanya@hp-notebook:~/Downloads/unified$ ./unified-parser नमस्ते 1 3 0 0
Segmentation fault (core dumped)
chaitanya@hp-notebook:~/Downloads/unified$ ./unified-parser नमस्ते 1 4 0 0
Segmentation fault (core dumped)

I kindly request to please help to solve this issue.

[Chaitanya Jadhav]

Hello all,

The problem was with GCC version. I installed gcc-4.8 and it worked.

In the make file change change CC = gcc to CC = gcc-4.8 (if you dont set gcc-4.8 as default comiler) and compile the program with make.

[IndicTTS]

Dear Chaitanya,

Thank you for your solution.We were using valgrind to solve the issue.

Regards,

TTS Group, IITM

[Chaitanya Jadhav]

Welcome Team !!

[Zhen Wei]

Hi Team & Chaitanya,

I also encountered this problem these days.

At the first stage, the segmentation fault appeared as follows:

then we delete '-std=c99' in Makefile, the fault appeared as follows:

then we fix a bug in unified.y at line 686, 39➡️38

and the segmentation fault disappeared!

But when I want to use unified-parser to process all the words in a corpus, the result shows that not all the words can pass the parser procedure, another faults “corrupted size vs. prev size” appeared as follows:

for example: प्रसिद्द , प्रस्तुत , राजनीति , प्रकृति ...

Since I cannot set breakpoint in .y or .l, I failed to use gdb to debug these faults. Please help!

I really hope that we can make this tool more feasible! 

Zhen Wei

[Speech and Music Technology Lab IIT Madras]

Dear Zhen Wei,

We are happy to help you. But we are unable to see the screenshot you've mentioned. Can you be bit more clearer ?

Thank you and Regards,

IndicTTS Team

IIT Madras

[Zhen Wei]

sorry for the failed images.

At the first stage, we got the segmentation fault, and then we delete '-std=c99' in Makefile, fixed a bug in unified.y at line 686, 39➡️38

Then, the segmentation fault disappeared!

But when I want to use unified-parser to process all the words in a corpus, the result shows that not all the words can pass the parser procedure, another faults “corrupted size vs. prev size” appeared as follows:

for example: प्रसिद्द , प्रस्तुत , राजनीति , प्रकृति ...

Besides, when we use absolute directory in unified.y (rootPath), some words (for example, श्रद्धा ) would have fault as follows: