Architecture: Speech Engine Services

Published: June 20, 2005

Architecture: Speech Engine Services

Speech Engine Services (SES) uses speech engines to process audio streams for both incoming and outgoing messages. The speech recognition (SR) engine interprets incoming speech from a user, and the playback engine produces outgoing speech streams to the user.

This section contains:

Service Access and Call Management

The way SES is accessed depends on the requesting device. For a telephony (voice-only) MSS system, Telephony Application Services (TAS) accesses SES. For a multimodal application, a client device (such as a PC or Windows Mobile-based Pocket PC (Pocket PC)) accesses SES through a TCP/IP network connection using Internet Explorer 6 or later, or Pocket Internet Explorer.

The SES lobby Web service (Lobby.asmx) manages all incoming recognition requests from telephony or multimodal clients, and returns the recognition result once the request has been processed.

When SES receives a request for speech recognition or speech output, it identifies an available instance of the appropriate engine. The engine instance then sets up the audio stream.

Single Server Deployment

See Also

Architecture: Telephony Application Services