Text-To-Speech Guide
© 2014 I6NET Solutions and Technologies, all rights reserved
Purpose
The VXI* VoiceXML browser integrates an HTTP client interface to connect to an HTTP text-to-speech (TTS) server. This allows for dynamically generated audio content with a text-to-speech engine. Most VoiceXML browsers have an MRCP (Media Resource Control Protocol) interface to access text-to-speech features. The advantage however, of using the HTTP connector is that the “speech” generated is cached by the VoiceXML Browser, and re-used the next time. The text is posted via an HTTP request then the server responds with a standard wav file. Users can use the TTS to generate menus prompts, for example, without need to purchase a lot of TTS licenses. The integrated TTS packages use PHP, so you need to have an Apache/PHP already installed on your server (MySQL is optional, it allows to produce “CDR” like reporting information to evaluate the TTS using).
Scope
- Mbrola TTS
- Nuance TTS
- Neospeech TTS
- Lumenvox TTS
- Ivona TTS
Usage
You can you two type of connections
- HTTP access
- MRCP access
HTTP
To use TTS by HTTP, you need to configure vxi.
In /etc/openvxi/client.cfg set POST
value in client.prompt.resource.0.method
key
client.prompt.resource.0.method VXIString POST
HTTP
To use TTS by MCRP, you need to configure vxi, and configure asterisk to use TTS with MRCP.
To configure VXI, set in /etc/openvxi/client.cfg
the ASTERISK
in client.prompt.resource.0.method
key
client.prompt.resource.0.method VXIString ASTERISK
To configure asterisk, update the file /etc/asterisk/mrcp.conf
[speech-loquendo-mrcp1] version = 1 server-ip = 192.168.100.164 server-port = 554 resource-location = media speechsynth = speechsynthesizer speechrecog = speechrecognizer rtp-ip = 192.168.100.134 ;rtp-ip = 192.168.1.2 ;rtp-ext-ip = auto rtp-port-min = 5100 rtp-port-max = 5999 playout-delay = 50 ;min-playout-delay = 20 max-playout-delay = 200 ptime = 20 codecs = PCMU PCMA L16/96/8000 rtcp = 1 rtcp-bye = 2 rtcp-tx-interval = 5000 rtcp-rx-resolution = 1000