*
* German Vocab for SynOhr Multisense KNX
*
SprachSchatz DE-2.505 (30-05-2017)
+New 
	New Standard - execution sound 
	Improved recognizer restart procedure
+Fixes 
	Buttonsounds stops playing audio output 

SprachSchatz DE-2.502 (14-12-2016)
+New 
	noise supresser with fast restart for better robustness of activation recognition
+Fixes 
	Fixes stop audio output 

SprachSchatz DE-2.404 (13-07-2016)
+New 
	new algorithm for gender specific wordprediction
	improved word discrimination algorithm
	customizable best fit stratgy 
	improved probability and confidence calculation 
	adapted Speech Quality Activation parameters for >MEDIUM
	adapted Speech Quality Command for all parameters 
+Fixes 
	n.a.	


SprachSchatz DE-2.100 (20-04-2016)
+New 
	Wordmodels adapted (improved learn sequence)
	Sensivity slightly changed.
	Changed _COLOR Wildcard: Direct Control of a 8 Bit H Telegramm (HSV Dimmuing) is possible
	   see wildcard.txt
	
+Fixes 
	Noise detection doesnot work
	If sounds are played sequentially, sometimes the audio output stops playing 
        till reboot
	Audio Volume adapted to new Firmware 1.082
	Reboot doensnot work from Menu with Firmware 1.082
	

SprachSchatz DE-2.090 (14-01-2016)
+New 
	Apdapted word models
	Improved peak detection
+ Fixes 
	n.a.

SprachSchatz DE-2.086 (11-12-2015)
+New 
	Adapted noise-modelling
	improved peak observer
+ Fixes 
        Noise stabilization didnot work properly

        
SprachSchatz DE-2.082 (26-11-2015)
+ Fixes 
        _LICHTER wasnot preconfigured properly.


SprachSchatz DE-2.081 (25-11-2015)
+New
	 New recognition algorithm for probability calculation. 
	    - Added more Gauss functions for better calculation resoultion (upto factor 3)
	    - Added new algorithm for decission trees (as in _PERCENT)
	    - Hand optimized code for many subroutines for faster operation
	 Improved robustness
	    - Improved detection of noise and talking sequences
	    - Added six algorithms for surpressing false magic word detection 
	    - Added extended algorithm for better detection of magic word. 
	 Added messages if parameter exceeds license model
	 Settings for "Sprechqualitaet Aktivierung" are changed. New setup will be necessary. 
	 New memory strategy for high performance calculation
	 Added flexible Framework for additional wildcard commands (needs SynOhrStudio 2.001)
	 Playing an audiofile will be interrupted, if playing a new audiofile is initiated

Fixes 
         In some circumstances the detection could de-rate. This is 
           now detected and automatically re-initalized (in less then 1ms)
         Wildcards didnot fit to new synohrcompiler 2.000
         Some noise as a coffee machine did trigger the magic word detection 


SprachSchatz DE-1.505 (15-06-2015)
+New
	 Added HILFE as magic word
	 This version is for heavy enviroment (people speaking, halls etc.)


SprachSchatz DE-1.504 (15-06-2015)
+New
	 Added HILFE as magic word
	 This version is for normal enviroment, 
	 children are recognized properly


SprachSchatz DE-1.502 (13-05-2015)
+New
	 Preparing recognizer for other language support
	 improved pre-compiling algorithm for offline trainig
	 added JAMES as magic word
	 improved recognition with echo signals 
	 
Fixes
	 Adjusted echo compensation 
	 Watchdog reboots when playing long audiofiles
	 
SprachSchatz DE-1.423 (20-03-2015)
+New
	 hand optimized critical routines
	 auto re-mount of SD Card for playing sounds
	 enlarged database with 20 new speakers 
	 new magic words 
	 adapted default parameters for speach recognition
	 autorecover for recording sessions (Trainer mode)
	 Sprachschatz compiled with high variance setting
	 
Fixes
	 touch sounds stop playing (II) 
	 system hang during recognition (IV)
	 triggering touch sensors during soft reboots
	 removed debug code/unused algorithms 
	 word ELF with better results

SprachSchatz DE-1.402 (10-12-2014)
+New
	 needs SynOhrStudio 1.102 or higher
	 rearranged best fit strategy with wrong probabilty path detection
	 improved magic word detection
	 added recognition watchdog 
	 added expert option for limitation of messages on the KNX-display
Fixes
	 touch sounds stop playing
	 system hang during operation (III)
	 slowly degrading recognution results during operation


SprachSchatz DE-1.308 (09-10-2014)
+New
	 recompiled word base for number recognition
	 adaptive echo surpression algorithm 
	 word propabilty caclulation with improved precision 
	 improved magic word detection
Fixes
	 state machine during recognition is not properly reset
	 system hang during operation (II)

SprachSchatz DE-1.303 (30-09-2014)
+ New    
	 recompiled and re-arranged word base
	 improved recognition discrimination for command sequences 
	 slightly changed t_ready sound
	 optimized processing time
	 improved junk-recognition algorithm for stabilization against noise
Fixes
	 system hang during operation

SprachSchatz DE-1.300 (03-09-2014)
+ New    complete rearanged noise surpression 
	 enhanced recognition performance for number discrimination
	 extremly improved recognition speed (if many commands are used)
	 enhanced robustness (+6 dB) in various situations whith sourrounding noise

SprachSchatz DE-1.253 (09-07-2014)
+ New    one new male speaker
	 changed messages on display for better setup procedure
	 Improved audio noise suppression 
Fixes
	 If touchkeys or central buttons are pressed fast, key sound will stop 

SprachSchatz DE-1.251 (25-03-2014)
+ New    New male speakers 
	 Adjusted phonetic table for noisy
	 Accelerated SynOhr Active message
Fixes
	 if an unkown file is chosen for playing, system halted

SprachSchatz DE-1.249 (07-03-2014)
+ New    Stabilization "Sprechqualtität Aktivierung" improved
	 Adjusted phonetic table
	 Output "not exact" if stabilization (see above) is acitve
	 
 Known  Under certain circumstances the "yellow light"-problem
	 can still occur. FW fix is necessary 

SprachSchatz DE-1.245 (19-02-2014)
+ New    Pscho-Acustic based Filterbank
	 Timing improvements ("yellow light"-problem)
	 Adjusted enviromental sounds
	 New male speaker for speech data base
	 Dynamic range adjustment lowered 
	 improved noise reduction if choosing TALKING and LOUD as options

SprachSchatz DE-1.242 (07-02-2014)
+ New:   Based on SprachKörper 1.235
         Added stabilzation method "talking"
         New DSP Software
         Microphone bias compensation
         Needs SynOhrStudio > 1.011
         
+ Fixes  Statemachine correction if recog. fails 
         
SprachSchatz DE-1.240 (03-02-2014)
+ New:   Based on SprachKörper 1.233
         Improved stabilzation for magic key word
         Improved range compensation
         New DSP Software
+ Fixes  Problems with key-File
         

SprachSchatz DE-1.233 (25-01-2014)
+ New:   Based on SprachKörper 1.221
         Avoids deadlock after reading license


+ Fixes: n.a.

SprachSchatz DE-1.229 (22-01-2014)
+ New:   Based on SprachKörper 1.224
         New arranged filterbank
         Improved search performance
                  
+ Fixes: n.a.

SprachSchatz DE-1.226 (20-01-2014)
+ New:   Based on SprachKörper 1.221
         Check for result differences in 1.222 
         
+ Fixes: n.a.

SprachSchatz DE-1.226 (20-01-2014)
+ New:   Based on SprachKörper 1.222
         Improved stabilization with longer words
         
+ Fixes: Overflow in recognition 

SprachSchatz DE-1.225 (10-01-2014)
+ New:   Based on SprachKörper 1.220
         New DSP FW for all DSPs
         More search space memory
       
+ Fixes: Improved search algo


SprachSchatz DE-1.224 (08-01-2014)
+ New:   Based on SprachKörper 1.220
         New DSP FW for all DSPs
         Improved female utterance detection
         Improved child utterance detection
         More search space memory
       
+ Fixes: Little Girls or women with high tone in voice are not recognized
         stability problems of speech recognition
         

SprachSchatz DE-1.221 (13-12-2013)
+ New:   Based on SprachKörper 1.218
         New DSP FW for all DSPs
         Improved female utterance detection
       
+ Fixes: KeyFile Problems 
              
SprachSchatz DE-1.219 (09-12-2013)
+ New:	
	
+ Fixes:
	KeyFile not properly read
	Recognition hangs on a click sound
	Soundfile for t_fail.wav is distorted
	
SprachSchatz DE-1.217 (15-11-2013)

+ New:	
	Initial Version
+ Fixes:
	Beta-Problems