
www.veear.eu
40
EasyVR 3
User Manual (1.0.11)
Error codes
Below the list of the most useful error codes that may be returned by training or recognizing commands.
03h
ERR_DATACOL_TOO_NOISY
too noisy
04h
ERR_DATACOL_TOO_SOFT
spoke too soft
05h
ERR_DATACOL_TOO_LOUD
spoke too loud
06h
ERR_DATACOL_TOO_SOON
spoke too soon
07h
ERR_DATACOL_TOO_CHOPPY
too many segments/too complex
11h
ERR_RECOG_FAIL
recognition failed
12h
ERR_RECOG_LOW_CONF
recognition result doubtful
13h
ERR_RECOG_MID_CONF
recognition result maybe
14h
ERR_RECOG_BAD_TEMPLATE
invalid SD/SV command stored in memory
17h
ERR_RECOG_DURATION
bad pattern durations
4Ah
ERR_SYNTH_BAD_VERSION
bad release number in speech file
4Eh
ERR_SYNTH_BAD_MSG
bad data in speech file or invalid compression
80h
ERR_NOT_A_WORD
recognized word is not in vocabulary
The first group of codes (03h
– 07h) is due to errors in the way of speaking to the EasyVR or disturbances in
the acquired audio signal that may depend on the surrounding environment.
The second group (11h
– 13h) indicates an insufficient score of the recognized word (from lowest to highest).
Acceptance of lower score res
ults may be allowed by lowering the “knob” or “level” settings, respectively for
built-in and custom commands (see CMD_KNOB and CMD_LEVEL).
A third group of codes (14h
– 17h) reports errors in the stored commands that may be due to memory
corruption. We suggest you check power level and connections, then erase all the commands in the faulty
group and train them again.
The fourth group (4Ah
– 4Eh) deals with errors in the compressed sound data, either because the wrong
version of the QuickSynthesis
TM
tool has been used to generate the sound table or because a not supported
compression scheme has been selected (or data is generically invalid).
The last code (80h) means that a word has been recognized that is not in the specified built-in sets. This is
due to how Speaker Independent recognition works and should be ignored.