Definitions
SymbolSet - a symbol set definition for a given language - https://github.com/stts-se/pronlex/tree/master/symbolset
Mapper - map transcriptions between different symbol sets for a given language - https://github.com/stts-se/pronlex/tree/master/symbolset/mapper
Converter - map transcriptions between different languages - https://github.com/stts-se/pronlex/tree/master/symbolset/converter
Validation - component for validating transcriptions partly based on a pre-defined symbolset - https://github.com/stts-se/pronlex/tree/master/validation

When we say "transcription converter", we mean the mapper and the converter. They are both heavily dependent on the symbol set package, so that package will probably go with them. Or? [TODO: decide]

In order to move the transcription converting calls from the pronlex API to a separate API, what do we do with the validation package? It is dependent on the symbol set (but no really the mapper or the converter). Can it stay in the pronlex repository? [TODO: decide]

If we keep the validation package in pronlex, and the command line tools for importing/converting/validation lexicon files and entries, the pronlex repository will have dependencies on the new package. Is this OK? [TODO: decide]

The symbolset has a dependency on lex.Entry but I think that could easily be removed if we put that logic somewhere else, and have the symbolset handle only single transcription strings (instead of full entries).

In the pronlex repository, the use of code related to the symbolset/mapper/converter packages are primarily used in the following components.

A complete list of usages:

Use of package symbolset:

• cmd/lexio/importLex/
• cmd/test_validator/
• cmd/validate_lex_file/
• dbapi/dbapi_test.go
• dbapi/validation_test.go
• lexserver/lexserver.go
• lexserver/mapper.go
• lexserver/symbolset.go
• lexserver/validation.go
• symbolset/converter/
• symbolset/mapper/
• validation/

Use of package symbolset/mapper:

• cmd/lexio/convert/CMU2WS/
• cmd/lexio/convert/csCzPhword2WS/
• cmd/lexio/convert/nbNoNST2WS/
• cmd/lexio/convert/svSeNST2WS/
• lexserver/mapper.go

Use of package symbolset/converter:

• lexserver/converter.go

Use of the mapper/converter service in the current version of wikispeech_mockup:

wikispeech_server/adapters/mapper_client.py:16:         self.base_url = "%s/mapper" % config.config.get("Services", "lexicon")
wikispeech_server/adapters/mapper_client.py:22:         url = "%s/%s/%s/%s" % (self.base_url, "maptable", self.from_symbol_set, self.to_symbol_set)
wikispeech_server/adapters/mapper_client.py:40:         url = "%s/%s/%s/%s/%s" % (self.base_url, "map", self.from_symbol_set, self.to_symbol_set, string)
wikispeech_server/adapters/marytts_adapter.py:21:       mapper_url = config.config.get("Services", "lexicon")
wikispeech_server/adapters/marytts_adapter.py:610:      url = mapper_url+"/mapper/map/%s/%s/%s" % (from_symbol_set, to_symbol_set, quote(trans))
wikispeech_server/adapters/marytts_adapter.py:649:      url = mapper_url+"/mapper/map/%s/%s/%s" % (from_symbol_set, to_symbol_set, quote(trans))

A subtask automatically inherits all project tags and subscribers. Just cleaning these up a bit.

@Sebastian_Berlin-WMSE @kalle @Lokal_Profil I found a note saying that we should inform you guys when this task is ready for testing from your side, so here we go: The updates have been pushed to master, and can be tested now (and docs updated).

The new repo: https://github.com/stts-se/symbolset

Our Wikispeech summary page: http://stts-se.github.io/wikispeech/