Corrects RobinASR (https://github.com/racai-ai/RobinASR) raw output by adding hyphenation and basic capitalization based on lists of named entities.
Runs as a Java REST API service.
java -cp deps\json-20200518.jar;deps\commons-lang-2.6.jar;RobinASRHyphenationCorrection.jar Server 8012 model2
java -cp deps/json-20200518.jar:deps/commons-lang-2.6.jar:RobinASRHyphenationCorrection.jar Server 8012 model2
While starting it will display the following:
Loading resources from [model2] All resources loaded Starting server on port 8012
After starting the program you can connect with a browser by using an URL similar to:
http://127.0.0.1:8012/correct?text=gheorghe%20%C8%99i%20ion%20sau%20dus%20la%20munte%20sau%20la%20mare (there is a single parameter called "text" receiving the text as an urlencoded value; for larger texts this can be send via POST)
The result is a JSON similar to the following:
{ "comments":[ "și [0] [18] [0] [0]", "ion [0] [53] [0] [0]", "s-au [2473] [40] [0] [0]", "la [0] [2914] [0] [0]", "sau [3] [119840] [0] [0]", "la [0] [5136] [0] [0]", "mare [0] [0] [16] [655443]" ], "text":"Gheorghe și Ion s-au dus la munte sau la mare", "status":"OK" }
The two relevant fields are "status" which should be "OK" and "text" which contains the corrected text. "comments" is an internal field describing how the model reached a decision. It's not relevant for the usage of the system.