I am having a very hard time getting anything to run in Sphinx3. I got everything to compile, following the directions in the online Tutorial. I also installed the an4 speech library, and ran all of the specified scripts. All I really need, however, is an example program like "Hello World" in Sphinx4. I've tried following the posts on this forum to get the live decode (or maybe I need continuous?) program to run using the following arguments (in a config file):
These arguments, however, just cause the program to throw an error after printing "Search Initialization". What am I missing? Is it the wrong dictionary? Do I need something else?
ALso, I would very much like to include a small grammar for my application. It shouldn't be much more complicated than that of Hello World. What additional arguments do I need to include in order to add a grammar file?
THank You,
Eric
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thanks for that. You've been a lot of help. I'll take a closer look at the continuous package then.
Regarding the sphinx_jsgf2fsg, I realized that I was not using the latest sphinxbase package and so did not have that file. However, the reason I wasn't using 0.4 is because it does not compile under visual studio 2005. In particular, it is missing the sphinx_jsg2fsg project file.... and without the associated dll, sphinxbase throws erros when compiling.
Eric
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Wouldn't you know it... but after working on this for two days, I finally got something to run an hour after posting a message for help. I ran the arguments in "args.an4.test.fsg" and livedecode ran, if not very well. Still haven't run anything else, but I have high hopes:)
This first success, however, brought up more questions.
1) Is there a utility for converting a sphinx4 grammar (jsgf?) to fsg? Earlier posts hinted at such a utility, but I cannot find it in the current release.
2) Do any of the existing demo programs demonstrate utterance detection? Pressing 'enter' at the beginning and end of each utterance nicely frames the speech query... but I don't think I'll have such info available in my app.
3) I also tried to run the continuous program without any luck. Sphinx3_continuous requests a 'ctrl_file' in addition to the raw file and a config file. What is a ctrl file?
Thanks Again,
Eric
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi! Yes, there is a utility for converting JSGF to FSG. It is called sphinx_jsgf2fsg, and it can be found in the latest sphinxbase package.
sphinx3_continuous is badly named since it isn't really equivalent to the pocketsphinx_continuous program. It only reads waveform files rather than accepting input from the audio device. However, it may be useful as an example of how to do endpoint detection.
The "control file" it expects is just a list of waveform files, or more specifically, filenames without the base directory and file extension.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I am having a very hard time getting anything to run in Sphinx3. I got everything to compile, following the directions in the online Tutorial. I also installed the an4 speech library, and ran all of the specified scripts. All I really need, however, is an example program like "Hello World" in Sphinx4. I've tried following the posts on this forum to get the live decode (or maybe I need continuous?) program to run using the following arguments (in a config file):
-hmm sphinx3/model/hmm/hub4_cd_continuous_8gau_1s_c_d_dd
-fdict sphinx3/model/lm/an4/filler.dict
-dict sphinx3/model/lm/an4/an4.dict
These arguments, however, just cause the program to throw an error after printing "Search Initialization". What am I missing? Is it the wrong dictionary? Do I need something else?
ALso, I would very much like to include a small grammar for my application. It shouldn't be much more complicated than that of Hello World. What additional arguments do I need to include in order to add a grammar file?
THank You,
Eric
Thanks for that. You've been a lot of help. I'll take a closer look at the continuous package then.
Regarding the sphinx_jsgf2fsg, I realized that I was not using the latest sphinxbase package and so did not have that file. However, the reason I wasn't using 0.4 is because it does not compile under visual studio 2005. In particular, it is missing the sphinx_jsg2fsg project file.... and without the associated dll, sphinxbase throws erros when compiling.
Eric
Wouldn't you know it... but after working on this for two days, I finally got something to run an hour after posting a message for help. I ran the arguments in "args.an4.test.fsg" and livedecode ran, if not very well. Still haven't run anything else, but I have high hopes:)
This first success, however, brought up more questions.
1) Is there a utility for converting a sphinx4 grammar (jsgf?) to fsg? Earlier posts hinted at such a utility, but I cannot find it in the current release.
2) Do any of the existing demo programs demonstrate utterance detection? Pressing 'enter' at the beginning and end of each utterance nicely frames the speech query... but I don't think I'll have such info available in my app.
3) I also tried to run the continuous program without any luck. Sphinx3_continuous requests a 'ctrl_file' in addition to the raw file and a config file. What is a ctrl file?
Thanks Again,
Eric
Hi! Yes, there is a utility for converting JSGF to FSG. It is called sphinx_jsgf2fsg, and it can be found in the latest sphinxbase package.
sphinx3_continuous is badly named since it isn't really equivalent to the pocketsphinx_continuous program. It only reads waveform files rather than accepting input from the audio device. However, it may be useful as an example of how to do endpoint detection.
The "control file" it expects is just a list of waveform files, or more specifically, filenames without the base directory and file extension.