Language Processing Server Support
U-Self Serve supports processing for both Media Resource Control Protocol (MRCP) and Automatic Speech Recognition (ASR).
U-Self Serve MRCP Server
U-Self Serve has an MRCP Speech processing component which can be used for IVR deployments. Enterprise contact center IVRs stream their audio via MRCP v2. A U-Self Serve MRCP server is available to process this audio stream and convert it to text in the chosen language and generate a voice response in the chosen language via an MRCP v2 stream.
The U-Self Serve MRCP engine includes the functionalities that follow:
Receive streaming audio from the IVR end point.
Stream the audio in real time to an ASR engine.
Receive the text transcription from the ASR engine.
Log the text transcription in the MRCP server for reference.
Return the text transcription to the IVR end point.
Generate a voice response as an MRCP v2 stream for a text input.
The ASR end points can be configured in the Uniphore ASR. Please contact Product team for the need to use any other 3rd party ASR engine.
The TTS Engine typically used is Amazon Polly. Please contact Product team for need to use any other TTS engines.
Uniphore ASR Integration
U-Self Serve can be integrated with the Uniphore ASR for speech to text conversion.
MRCP v2 is the only streaming audio supported. The Uniphore ASR is hosted separately in an end point as part of the deployment (On premise or Cloud). The latency of the ASR response depends on the compute power assigned to the Uniphore ASR instance.