Hello World Decoder QuickStart Guide Pt 4

From Sphinx

If you have followed the QuickStart Guide Part 1 and 2 you have everything in place to test out using the live decoder. You can see the Sphinx3 decoder in real time. The executable is called sphinx3_livedecode. When you build and install the Sphinx3 distribution it should be on your system, on my system it is in /usr/bin/sphinx3_livedecode.

To invoke it you run it from the the directory you created for Hellow World Decode Project Part2, e.g. $SPHINX_ROOT/hello_world_pt2. The sphinx3_livedecode will take the same config file that you created for the sphinx3_liverpretend. So to run it you enter the command line:

user@machine:/usr/src/sphinx/hello_world_pt2> sphinx3_livedecode cfg_file

At this point you should get the decoder initialize with lots of debug output and give you the prompt to "ENTER" to start recording:

......
INFO: s3_decode.c(259): Input data will NOT be byte swapped
INFO: s3_decode.c(264): Partial hypothesis WILL be dumped
press ENTER to start recording

At this point you should have your microphone ready so that you can give the decoder audio to process and you should see the live hypothesis streaming on the console. Here is a sample of what typical output you should expect:

press ENTER to finish recording
Warning: Could not find Capture element
Partial hypothesis:

Partial hypothesis:

Partial hypothesis:
HELLO
Partial hypothesis:
HELLO
Partial hypothesis:
HELLO
.........
Partial hypothesis:
HELLO
Partial hypothesis:
HELLO PLEASE
Partial hypothesis:
HELLO PLEASE
.........
Partial hypothesis:
HELLO PLEASE YELLOW
...........
Partial hypothesis:
HELLO PLEASE YELLOW
Hypothesis:
HELLO PLEASE YELLOW
user@machine:/usr/src/sphinx/hello_world_pt2>

Note that you need to press the ENTER key a second time to terminate the session and you see the final hypothesis at the end.