The Conversation Streams API allows streaming the agent and/or caller audio in real-time.
The Conversation Streams API opens the Talkdesk Voice platform so that customers and partners can have secure access to the raw audio stream of any phone call, for use cases like customer authentication via biometrics, real-time voice analytics and other voice analysis, or voice processing.
Talkdesk Conversation Streams can stream bidirectional audio, allowing it to leverage additional use cases like third-party voice virtual agents.
As a real life example, consider the scenario where a company is working with a service for building conversational interfaces into any app, using voice and text, which powers a virtual assistant.
The Conversation Streams API follows these steps:
1 - Receiving audio stream from the Talkdesk Voice platform.
2 - Sending the audio stream to a voice activity detection component to determine voice in audio.
3 - Starting streaming the user data to the service for building conversational interfaces when voice is detected.
4 - Stopping streaming the user data to the service for building conversational interfaces when silence is detected.
5 - Updating the ongoing Talkdesk call based on the response from the service for building conversational interfaces.
Additionally, a set of commands can be sent back to the Talkdesk Voice platform to:
- Update the stream with the service for building conversational interfaces' response.
- Escalate the call to an agent, with some limited parameters.
- Terminate the call.
The Conversation Streams API is only available for customers enrolled in Talkdesk's Early Access (EA) program. To enter the EA program, contact a Customer Success Manager.
The Conversation Streams has been in EA since January 2022.
Updated about 1 month ago