Most users will interface with the LumenVox ASR and TTS software using the Media Server, an application that usually runs as a background process (also known as a service or a daemon). The Media Server provides functionality to support versions 1 and 2 of the Media Resource Control Protocol. This is a standards–based method of interacting with the Speech Engine and TTS Server over a network. The Media Server handles both Real Time Streaming Protocol (RTSP) and Session Initiation Protocol (SIP) dialogues to initiate new connections from client applications to make use of LumenVox media resources. As part of these sessions, the Media Server accepts and sends Real Time Protocol (RTP) traffic containing audio streams meant for the Speech Engine or synthesized audio coming from the TTS Server.

The Media Server is available for both Windows and Linux, and makes use of the LumenVox API (see below for more information). Because the internal communication between LumenVox processes is TCP/IP–based, the Media Server may run on separate machines from the Speech Engine or TTS Server (though they may all reside on the same machine).

Included with the Media Server installation is a real–time Media Server Monitor, which provides users a graphical interface for viewing live MRCP sessions. This is a Windows–only tool.

The LumenVox Speech Engine is our flagship speech recognition application. It is available on Linux and Windows, and supports a number of languages. Its technology is based on industry standards and brings quality speech recognition to anyone who wants to speech enable an application.

The LumenVox Speech Engine is used in telephony applications, such as an IVR. It can also be used for any computerized command and control application, such as a voice dialer, order taker, or even to control simple actions on a PC. The Speech Engine is not dictation software and it does not transcribe large audio files. It is used for applications where users speak short, discrete phrases.

The LumenVox Speech Tuner is designed to perform tuning and transcription, as well as instant parameter, grammar, and version upgrade testing of any speech recognition application. It reduces the work of your post–deployment application revisions and allows you to bring tuning in-house, avoiding costly professional service fees.

• Analyze each stage in the call process.
• Transcribe audio data, make pinpoint adjustments, and immediately measure the effects on the overall performance.
• Test design and development decisions of new applications, using data from deployed applications.

LumenVox's Answering Machine Detection (AMD) technology delivers outbound messages by detecting answering machine and voicemail tones. The AMD is tested at 94%* accuracy. Our tone–based detection is a significant innovation compared to many other energy–based answering machine detection technologies. The LumenVox AMD actually listens to tones

facilitating more reliable and highly accurate message delivery. Since LumenVox is listening for the beep, your outbound dialing application won't know it's connected to a voicemail system at the beginning of a call. With higher tone–detection accuracy your message sounds more professional because it starts precisely where it should.

The LumenVox Text–to–Speech Server (TTS) provides text–to–speech synthesis, turning written text into spoken speech. It supports the Speech Synthesis Markup Language (SSML), a way to format text within XML to control the pronunciation of words, but it can also accept

plain text in order to perform synthesis. As with the LumenVox Speech Engine, the primary method of interacting with the TTS Server is through the Media Server, in which case synthesized audio is provided back to the client application as an RTP stream.

All products require a valid LumenVox license to use. The License Server provides an intelligent method of managing and distributing licenses. As with all communication between LumenVox applications, license requests are TCP/IP–based, allowing License Servers to be on different machines from the applications requesting licenses. Applications and License Servers can have many–to–many relationships, allowing a cluster of machines to each communicate with multiple License Servers for redundancy and load balancing reasons.

A License Manager utility is provided for users to interact with the License Server. This utility is either a graphical interface (on Windows) or a console application that accepts command–line parameters (on Linux). The License Manager utility is used to install or uninstall licenses, and to provide information about the licenses that are installed on a given machine.

The License Server itself is available for both Windows and Linux.

Though many users will choose to interface with LumenVox software through MRCP and the Media Server, LumenVox does provide a public API and the requisite libraries (DLL files on Windows and shared object files on Linux) for users to build their own applications around. The LumenVox libraries provide C and C++ functions that can be used to achieve all of the functionality

included in the Media Server and more. For instance, while the Media Server assumes a real time audio stream, the API includes functions to load audio from files on disk.

The API is fully documented and supported, and is available for both Windows and Linux.