Resource Packs

The following resource packs are available for IDOL Speech Server operations.

Pack Description Configuration requirement
Audio Security Pack Contains the EVENTS pack required by the audiosec module. Configure in the [Resources] section of the speechserver-tasks.cfg file.

In individual tasks that use the audiosec module, specify the EVENTS pack as the value of the Model configuration parameter.
Language Pack (various languages) Contain the files necessary for IDOL Speech Server to process various languages. They are available for several languages, as well as regional variations of the same language (for example, U.S. English and British English). Available language packs lists the available language packs.

The language packs also contain files that are adapted for phonetic phrase search.
Configure in the [Resources] section of the speechserver-tasks.cfg file. The configuration file must contain separate entries for the standard and phonetic phrase search versions of a language pack. The standard version is a lang resource type; the phonetic phrase match version is a langvt resource type and adds a -pm suffix to the pack name. For example, ENUS is the standard language pack, and langvt:ENUS-pm is the phonetic phrase search version.
Telephony Pack (various languages) Contain the same files as the standard language packs, except that these are adapted for processing 8 kHz audio instead of 16 kHz. Configure in the [Resources] section of the speechserver-tasks.cfg file. The configuration file must contain separate entries for the standard and phonetic phrase search versions of a language pack. The standard version is a lang resource type; the phonetic phrase match version is a langvt resource type and adds a -pm suffix to the pack name. For example, ENUS-tel is the standard language pack, and langvt:ENUS-tel-pm is the phonetic phrase search version.
Language ID Pack Contains the base classifier pack, which provides language classifiers for 20 languages (for a list of included languages, see Create Your Own Language Classifiers).

Also contains the SYLS language pack required to create language identification feature files from 16 kHz audio data.
For the base classifier pack, set the LangList parameter in the langid and lidoptimizer modules to a subset of languages to use from the pack.

For the SYLS language pack:
  • Configure in the[Resources] section of the speechserver-tasks.cfg file.
  • In individual tasks that use the lidfeature module, specify the pack as the value of the Lang configuration parameter.
Language ID Telephony Contains the SYLS language pack required to create language identification feature files from 8 kHz audio data. Configure in the [Resources] section.
In individual tasks that use the lidfeature module, specify the pack as the value of the Lang configuration parameter.
Speaker ID Pack Contains 8 kHz and 16 kHz versions of base .atf, .ian, .sig and .ast files.

The .atf file is used as a base template when training, optimizing, and packaging new audio template files (such as speaker templates).

The .ian file is used during audio feature frame normalization.

The .ast and .sig files are used during speaker segmentation.
Configure in the [Resources] section.

In individual tasks that use the normalizer, speakerid, audiotemplatetrain, audiotemplatedevel, audiotemplatescore, or audiotemplateedit modules, specify the speaker ID pack as the value of the SpkIdBasePack configuration parameter if required.

Available language packs

Language Language pack
Arabic
  • Gulf Dialect: ARGU
  • Modern Standard Arabic: ARMSA
Catalan
  • CAES
Czech
  • CSCZ
Danish
  • DADK
Dutch
  • NLNL
English
  • Australian English: ENAU
  • British English: ENUK
  • Canadian English: ENCA
  • Irish English: ENIE
  • Singaporean English: ENSG
  • U.S. English: ENUS
Flemish
  • NLBE
French
  • Canadian French: FRCA
  • French: FRFR
German
  • DEDE
Greek
  • ELGR

Hebrew

  • HBIL
Hindi
  • HIIN
Hungarian
  • HUHU
Italian
  • ITIT
Japanese
  • JAJP
Korean
  • KOKR
Mandarin
  • ZHCN
Persian
  • FAIR
Polish
  • PLPL
Portuguese
  • Portuguese (Portugal): PTPT
  • Brazilian Portuguese: PTBR
Romanian
  • RORO
Russian
  • RURU
Slovak
  • SKSK
Spanish
  • Castilian Spanish: ESES
  • Latin American Spanish: ESLA
  • North American Spanish: ESUS
Swedish
  • SVSE
Turkish
  • TRTR
Welsh
  • CYUK
NOTE:

For certain languages only the broadband or telephony pack is available, not both.


_HP_HTML5_bannerTitle.htm