This is a directory used by Barbara Hohlt and Steve Czerwinski
for the TranscoderService which is part of ICEBERG pre-Release 0.0.
It can be found in the ICEBERG download ==>
	http://iceberg.cs.berkeley.edu/release 

Below are instructions for installing the service.

Pre-Installation Instructions
1. The TranscoderService executes as a Ninja Service in
	a Ninja iSpace. Download and install Ninja 1.5 at
		http://ninja.cs/software.html
2. Install GSM toast
	http://kbs.cs.tu-berlin.de/~jutta/toast.html
3. Install sox
	ftp://sunsite.unc.edu/pub/Linux/apps/sound/convert/
			

4. The TranscoderService requires the installation
	of IBM ViaVoice and ViaVoiceOutloud, 
	applying some modifications (viavoice_mods.tar),
	and editing some paths.
a) Install ViaVoice SDK and Runtime
	http://www-4.ibm.com/software/speech/dev/sdk_linux.html
	save some files (optional)
	
	// custom audio library
	cd ViaVoice/samples/audio
	cp audbase.c audbase.c.save

	// dictation grammar
	cd ViaVoice/samples/dwplus
	cp Makefile Makefile.save
	cp dw.c dw.c.save

	// rule grammar
	cd ViaVoice/samples/gramsam
	cp Makefile Makefile.save
	cp gramasc.c gramasc.save

b) Install ViaVoiceOutloud SDK and Runtime
	http://www-4.ibm.com/software/speech/dev/ttssdk_linux.html
	save some files (optional)
	// synthesizer
	cd ViaVoiceOutloud/samples/cmdlinespeak
	cp makefile.linux makefile.linux.save


c)	The modicications in the .tar file create a ViaVoice custom
	audio library, audfile, which supports file I/O to and from
	the speech recognizer and synthesizer.
	cd to directory that contains ViaVoice and ViaVoiceOutloud directories 
		(i.e. that is the parent directory of the two)
	tar -xvf viavoice_mods.tar 
	
d)	Change paths in files if necessary and make
 	Note: If you didn't install ViaVoice in /usr/lib you will need
   	to change some paths. In any case you will need to run make.

	cd ViaVoice/samples/audio/
	Change path if necessary in audbase.c::AudioCreate()
	cd ViaVoice/samples/audio/audfile
	make

	cd ViaVoice/samples/dwplus
	Change path if necessary in runascdict 
	make

	cd ViaVoice/samples/gramsam
	Change path if necessary in rungramsam
	make

	cd ViaVoiceOutloud/samples/cmdlinespeak
	Change path if necessary in runcmdlinespeakfileout (it's
		possible the path to eci.ini is different, check 
		your .save file)
	make -f makefile.linux


Installation of TranscoderService
    
1. mkdir /tmp/analout

2. Change paths if necessary in run* files for:
	iceberg/transcoder/run*
	iceberg/transcoder/analyzer/runaudioanal

3.  Set properties in .ninja/config.  You will need to use your own paths
	where necessary. The property, external.transcoder.address, is 
	the host address and iSpace port number of the TranscoderService. 
	If you are using a default iSpace port them don't add the ':xxxx'.
	
# EXAMPLE: Needed for Transcoder
iceberg.transcoder.address myhost.cs.berkeley.edu:3344
external.transcoder.recognizer /usr/lib/ViaVoice/samples/dwplus/runascdict
external.transcoder.recocmd /usr/lib/ViaVoice/samples/gramsam/rungramasc
external.transcoder.synthesizer /usr/lib/ViaVoiceOutloud/samples/cmdlinespeak/runcmdlinespeakfileout
external.transcoder.gsm		/bin/rungsm
external.transcoder.sox22   /bin/runsox22
external.transcoder.append  /bin/runappend
external.transcoder.cleanup	/bin/runcleanup
external.transcoder.audioanal /transcoder/analyzer/runaudioanal

4.	run make in the analyzer directory

5. edit and run make in the transcoder directory

6. If you want to use the Transcoder without installing the Mediamanager, 
just comment out any references to MediaManager and its classes and it
will work fine. 

7. Note: References to the Ninja SDS Service have been commented out.
If you would like to use SDS with the Transcoder Service for
service discovery, just remove the comments and make.