New Features

• Initialized the AWS SDK in a certain way to use a fixed pool of threads for execution of HTTP requests. The default behavior was to spawn a thread per request, which is not efficient enough for processing hundreds of concurrent requests.
• An AWS region can now be specified from the configuration.

Fixed Problems

• If 'bypass-ssml' is set to false, then the SSML content is recomposed to strip off the 'voice' element before sending the content to the service.
• All the internal common class names are declared under a unique namespace not to cause a conflict with other speech synthesis plugins being loaded in to the same instance of UniMRCP server.
• If 'save-waveforms' was set to 'true' while 'save-records' was set to 'false', the header of saved wave files was not properly composed.

Configuration Parameters

• Added new configuration attributes 'init-sdk' and 'shutdown-sdk' to the element 'umspolly'.
• Added a new configuration attribute 'thread-pool-size' to the element 'synth-settings'.
• Added a new configuration attribute 'region' to the element 'synth-settings'.

Miscellaneous

• Updated the Usage Guide to reflect the changes introduced in this release.