julius icon indicating copy to clipboard operation
julius copied to clipboard

Published 20 hours ago • verified publisher icon julius-speech

segfaults on linux

Open lsth opened this issue 5 years ago • 5 comments

julius compiles fine, but when I try to execute the example from the README

../julius/julius -C julius.jconf -dnnconf dnn.jconf

it segfaults:

### Recognition: 2nd pass (RL heuristic best-first) Segmentation fault (core dumped)

see full build & run logfile attached.

any idea what I am doing wrong?

thanks! & best lars

OS: opensuse linux 15.1 julius: bff4dfedffa78b1afb76d4a4126b8a009eff1cfe log of build & run: mylog.txt log (with trace): mylog-strace.txt

lsth avatar Nov 27 '19 21:11 lsth

I'm having the exact same segmentation fault in the exact same location as described above on Ubuntu 18.04.2 LTS when using julius with the English DNN models as specified on the README page. Any ideas what is wrong? Do we need to use GMM models with this version of julius?

asrivast13 avatar Jan 15 '20 19:01 asrivast13

Did you try with other Wav file?

Amit Srivastava [email protected] ezt írta (időpont: 2020. jan. 15., Sze 20:45):

I'm having the exact same segmentation fault in the exact same location as described above on Ubuntu 18.04.2 LTS when using julius with the English DNN models as specified on the README page. Any ideas what is wrong? Do we need to use GMM models with this version of julius?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/julius-speech/julius/issues/123?email_source=notifications&email_token=AHNHTWQ655CDCP7MQFH3APTQ55RUFA5CNFSM4JSMPZ72YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJBSHUI#issuecomment-574825425, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHNHTWWCY7NNH7UBYCUIOGDQ55RUFANCNFSM4JSMPZ7Q .

Estalhun avatar Jan 16 '20 05:01 Estalhun

Getting this same segfault on the 2nd <<< please speak >>> microphone input on the 2nd pass

Wesxdz avatar May 28 '20 06:05 Wesxdz

Just to clarify what may not be immediately obvious https://sourceforge.net/projects/juliusmodels/files/ says:

  1. ENVR-v5.4.Dnn.Bin.zip DNN English speech models (262K dictionary, 32 bit LM)

To use these models you need a modified version of Julius from: https://github.com/palles77/julius

SO DO NOT USE THIS ZIP FILE, SELECT ANOTHER ONE THAT IS COMPATIBLE WITH THE PROJECT THIS COMMENT IS POSTED AGAINST https://github.com/julius-speech/julius

dlmiles avatar Sep 07 '20 06:09 dlmiles

The document in the zip is old. the diff of https://github.com/palles77/julius has been merged so you can run it with latest Julius, not the modified version.

Please tell me if you still have error with the latest version.

LeeAkinobu avatar Sep 08 '20 08:09 LeeAkinobu