text to speech

[TwistedFister]

Vicibox 7.0.3 from .iso | Vicidial 2.14-579a Build 161128-1746 | Asterisk 11.22.0-vici | CLuster setup: 1 web 1 DB 8 telephony | No Digium/Sangoma Hardware | No Extra Software After Installation

desired scenario:
a person that we have called before calls in and when they call in they hear a greeting (hi hello whatever) after that they hear the details of THEIR info from our database to further personalize the call and then given the option to select routing from IVR. We sell advertising for used cars for sale so the work flow would be that we have called you but you didnt answer so you call us back. When you call us back it sees the phone number and says "hello i see you are returning the call regarding your 2015 Ford F150.. would you like to speak to a...."

is this remotely doable?

if so message me with a price estimate to create and/or configure.

Thanks!

[mflorell]

Yes, it is doable, we have done similar projects before.

Do you have an estimate on the number of concurrent channels that you will need to handle with this application?

Do you have any current Cepstral text-to-speech licenses?

Could you describe you current VICIdial cluster setup?

[TwistedFister]

not sure on the concurrent channels but we average about 9000-10000 inbound calls per day with 85 agents taking them with DROP rate of 15% all spanned over 9 hours.

we don't have any Cepstral licences as of yet but we have no problem purchasing.

Our vicibox cluster is 1 web 1 DB w/ slave to run our own reporting, FTP for recordings, 7 dialing nodes and 1 node dedicated to inbound calls and voicemail.
HARDWARE:

all are Dell 1950s w/dual zeons 32GB ram 64GB on the DB and all are on Samsung SSD's.

Dialing:
2 blended campaigns running in ratio with 42 agents each.. Roughly 90,000 outbound calls a day using AC-CID 9000-10000 inbound calls a day.

[mflorell]

The good news is that if all of your inbound calls are coming in on one server, then you will only need the Cepstral licenses on that one server.

As for the number of concurrent inbound calls, that isn't as easy to figure out. A good starting place is to run the Inbound Report and look at the highest "INITIAL QUEUE POSITION". Then look at the Max System Stats and see what the maximum concurrent inbound calls are over the same time period. The number of concurrent channels you'll need are probably going to be somewhere in the middle. Of course, we recommend buying more licenses than you think you'll need because if it does run over, then no audio will be generated.

The next step would be to write up a full call flow for how you want the Text-to-Speech(TTS) portion to work. This will need to include all of the variables that the TTS application will need to access and what the possible options are for the customer to select as they go through the Call Menu tree for this.