Andrew's Asterisk Stuff

VM next message mods:

I have an old PBX VM system that allowed me to wrap around during voicemail message selection. At the first message I am able to select "previous" and have it jump back to the last message (so I could keep playing them in reverse order). Similar for the last message, it allows me to select "next" and it will jump to the first message. This does not force a change to the order of playback it just allows you to go in a circle to get the message you want quickly.

This new feature is enabled by default but can be disabled in the voicemail.conf file. The system knows if there are no messages, or just one, and does not play the prompts for prev/next since it's pointless.

This patch is for Branch 1.4.

Included english audio files are not correctly licensed for distribution with asterisk and should be re-recorded to match the other asterisk prompts anyway.

vm-goingtofirst.wav "Going to the first message."
vm-goingtolast.wav "Going to the last message."

Copy WAV files to /var/lib/asterisk/sounds (and transcode as needed).
Copy patch to main asterisk source directory and run patch -p0 <vm.patch and rebuild.

TTS Mods:

I now use Cepstral as a cost effective (but not free) solution.

Cepstral is easy to use, supports Linux and Solairs, works well, sounds very good, and has a choice of voices including Allison (who records the packaged prompts for Asterisk). They offer full demo downloads so you can test a working setup before you pay for it.

At their current prices, it's $30 for the each voice and a single queued FIFO license (no matter how may voices you have paid for). For additional "ports" (concurrent use) it's $50/each in addition to the voices. You have to pay for at least 2 total concurrent uses ($100) to get beyond the FIFO setup.

There are at least three ways to use Cepstral with Asterisk:

• Direct realtime text to speech to the current channel: There are several applications but the one that works the best is app_swift (by Will Orton). This uses some CPU time and memory to generate the speech output each time. If you have a FIFO license there will be a delay if more than one TTS session is needed at the same time.
• Static audio files: You create audio files using the swift application included with the package. It allows you to select any voice options and output the audio to a file (wave or ulaw). There is no additional CPU time used for playback, but everything will have to be manually recorded.
• Automatic cached static audio files: I wrote the appliction named app_swiftcache that shares app_swift's config file and automaticly builds speech audio sound files.

Aplications

• app_swift by Will Orton (currently version 0.9.1)
• app_swiftcache by Andrew Lindh (version 1, tested only on Linux)

app_swiftcache

Config options include:

• directory is where the cached audio files should saved. If it's not listed or valid then "/tmp" will be used.
• generalinfofile sets the name of the single text file were the voice name and text of each recording will be stored, defauilts to speech.txt
• individualinfofile if set to yes (default) will also save the same voice name and speech text in the same file name as the audio (with .txt extension).
• forcerebuild if set will always overwrite the audio sound file even if it exists.
• autorebuild if set to more than zero will rebuild the cached sound file if it is more than X hours old.

Examples

[cache]
directory=/var/lib/asterisk/swiftcache
generalinfofile=speech.txt
individualinfofile=yes
forcerebuild=no
autorebuild=0

exten => 440,n,Swift(Allison likes bigfoot.)

exten => 441,n,SwiftCache(Allison likes bigfoot.)
exten => 441,n,Playback(${SWIFTCACHE_FILE})

Transfer Mod:

After an attended SIP transfer is completed play a beep.
This is a standard feature of 1.6

Patch file for 1.4 (branch): transfer_beep_14.patch.gz
Copy patch file to your main asterisk source directory
run patch in Asterisk directory: patch -p0 <transfer_beep_14.patch
rebuild and install

I just add this to the [global] section of the extensions config file: ATTENDED_TRANSFER_COMPLETE_SOUND=beep

FAX Mods:

Install additional packages needed (debian examples: apt-get install autoconf)
Download SPANDSP 0.0.4 (currently version 0.0.4pre18)
Configure SPANDSP: ./configure
Install SPANDSP: make install
Install apps in asterisk apps directory.
modified: apps/app_rxfax.c
modified: apps/app_txfax.c
updated patch: fax.patch (for branch 1.4 04-Dec-2007)
Copy patch file to your main asterisk source directory
run patch in Asterisk directory: patch -p0 <fax.patch
remake the asterisk configure files: bootstrap.sh
enable the applications from menuselect: make menuselect
Build Asterisk: make

Dialplan example:

[default]
exten => 8605551212,1,Macro(rxfax-email,someone@somewhere.nett)

[macro-rxfax-email]
exten => s,1,Answer()
exten => s,n,Set(TIMESTAMP=${STRFTIME(,EST5EDT,%Y%m%d-%H%M%S)})
exten => s,n,Set(FILEID=${TIMESTAMP}_${UNIQUEID})
exten => s,n,Set(FAXPAGES=0)
exten => s,n,RxFAX(/tmp/fax-${FILEID}.tif)
exten => s,n,Hangup()

exten => h,1,NoOp("Finish with RxFax")
exten => h,n,GotoIf(${FAXPAGES}>0?process:done)
exten => h,n(process),system(/usr/bin/tiff2pdf -z -n -f /tmp/fax-${FILEID}.tif | /usr/bin/mime-construct --header "From: FAX " --to "${ARG1}" --subject "FAX from ${REMOTESTATIONID} - ${FAXPAGES} Pages" --string "FAX PDF Attached" --type application/pdf --attachment "fax.pdf" --file - --output | /usr/lib/sendmail -F"FAX" -f"fax@somwhere.nett" "${ARG1}")
exten => h,n(done),system(/bin/rm /tmp/fax-${FILEID}.tif)
exten => h,n,Hangup()

Remote called Party ID:

My current patch I use for Branch 1.4 remote_id_branch.patch (updated 11-Nov-2007)
Copy patch file to your main asterisk source directory
run patch in Asterisk directory: patch -p0 <remote_id_branch.patch
enable the dialplan function from menuselect: make menuselect
Build Asterisk: make

G.722 Mods:

SVN rev 114550 of 1.4 is needed for G.722 support.
Upgrade to Asterisk 1.6 and get official support.

Notes:

• Most of these issues may be internal code problems that will not be fixed in Asterisk 1.4
• This is the free G.722 codec, not the G.722.1 or G.722.2 codec.
  
• I have used it with the Polycom SoundPoint IP 650 (550 also supports it).
  The Polycom is VERY LOUD when using G.722 with Asterisk, but seems ok between polycoms
  
• The Grandstream GXP-2000 seems work partly with current fimware (1.1.4.25). The audio quality and volume at the phone are good. But the speaking/trasmit audio seems to have issues. Some times it's muted (other party hears nothing), choppy (sounds like a lot of packet loss), or sounds like it's underwater (wrong timing/duplicate packets).
  

Codec Negotiation: