voicevirtualagent.proto

/*
This proto file has the definition of RPC, request and response structures required to use BYoVA(Bring your own Virtual Agent)
These definitions will be used for communication between client and VA(server) over gRPC
*/
syntax = "proto3";
package com.cisco.wcc.ccai.media.v1;

import "com/cisco/wcc/ccai/media/v1/common/byova_common.proto";
import "google/protobuf/timestamp.proto";
import "google/protobuf/struct.proto";

/*
 Represents the Request format to be sent by the client to communicate with VA(server)
*/
message VoiceVARequest {
  // Conversation id - mapped to call id/interaction id
  string conversation_id = 1; // Mandatory for all request

  // Customer organization ID.
  string customer_org_id = 2; // Mandatory for all request

  // ID of the virtual agent that must be invoked. This will be provided by vendor using List virtual agent response
  string virtual_agent_id = 3; // Optional

  // Indicates whether partial responses from the virtual agent are allowed.
  bool allow_partial_responses = 4;

  // mapped to ccai config created via Control hub
  string vendor_specific_config = 5;

  // Type of input supported in the request.
  // One of voice, dtmf, or input event must be included.
  oneof voice_va_input_type { // Mandatory for all request
    // The voice input from the caller(caller audio)
    VoiceInput audio_input = 6;

    //DTMF events during the call.
    DTMFInputs dtmf_input = 7;

    //Input events, such as call start, call end, no input, etc.
    EventInput event_input = 8;
  }
  //Optional:Map to capture any additional or miscellaneous info.For future use and not supported currently
  map<string, string> additional_info = 9;
}

/*
 Represents the voice input object(captures information related to caller audio)
*/
message VoiceInput {
  // The raw audio bytes for the caller's audio stream.
  bytes caller_audio = 1; // Mandatory for all request

  // Encoding format of the audio data.
  enum VoiceEncoding {
    UNSPECIFIED_FORMAT = 0; // none
    LINEAR16_FORMAT = 1; // 16-bit linear PCM.
    MULAW_FORMAT = 2;  // G.711 mu-law.
    ALAW_FORMAT = 3; // G.711 A-law.
  }
  VoiceEncoding encoding = 2; // Value of selected encoding format.Mandatory for all request

  // Sampling rate of the input audio in Hertz.
  int32 sample_rate_hertz = 3; // Mandatory for all request

  // Start timestamp of when the audio data was captured.
  google.protobuf.Timestamp audio_timestamp = 4; // Mandatory for all request

  // Language code of the caller, e.g., 'en-US'.
  string language_code = 5; // Mandatory for all request

  // Indicates if the audio content represents a single utterance.Default value is false
  // This is sent by client if support of single utterance is required
  bool is_single_utterance = 6;
}

// Represents the output of the virtual agent, which includes response audio, events, and configurations.
// This is sent by VA(server) after processing the request.
message VoiceVAResponse {
  // List of prompt Voice responses to be played by the caller.
  repeated Prompt prompts = 1;

  // Output events from the virtual agent, such as session end or transfer to human agent.
  //As of now only one event will be supported and first event will be considered.
  repeated OutputEvent output_events = 2;

  // Indicates whether the next input from the client is to be considered sensitive (e.g., for PCI compliance).
  // Default value is false
  bool input_sensitive = 3;

  // Input mode for next input(voice ,dtmf etc)
  VoiceVAInputMode input_mode = 4;

  // Speech timers and DTMF configuration for handling input.
  // This tells client what is expected by VA(server) in terms of configurations.
  InputHandlingConfig input_handling_config = 5; // Mandatory for all request

  // Optional. Final transcript of entire session,
  // typically included in last response.
  // Transcripts included in intermediate responses are ignored.
  TextContent session_transcript = 6;

  // Optional. Summary of the session, included in the last response.
  // Summary included in intermediate responses are ignored.
  // SSML does not make sense for summary.
  // Using TextContent so that language code can be used.
  TextContent session_summary = 7;

  //Type of response VA is sending(Partial/Final/Chunk)
  ResponseType response_type = 8;

  enum ResponseType {
    FINAL = 0;               // No more response expected
    PARTIAL = 1;             // Expect more response from server
    CHUNK = 2;               // To enable the streaming flow.
  }
}

// Describes the prompt (Voice or text) to be played to the caller.
message Prompt {
  string text = 1; // Text of the prompt (if available).
  string audio_uri = 2; // URI of the audio to be played.
  bytes audio_content = 3; // Raw audio content.
  // Whether the caller can barge in before the prompt is completely played out
  bool is_barge_in_enabled = 4;
}

// Type of input expected from user
enum VoiceVAInputMode {
  INPUT_VOICE_MODE_UNSPECIFIED = 0; //unspecified input
  INPUT_VOICE = 1;   //voice input(caller audio)
  INPUT_EVENT_DTMF = 2; //event dtmf input(dtmf events)
  INPUT_VOICE_DTMF = 3;  //voice dtmf input(voice dtmf)
}

// Service definition for the Voice Virtual Agent gRPC API.
service VoiceVirtualAgent {

  // Bidirectional streaming RPC to send and receive caller audio, DTMF,
  // or input events.
  rpc ProcessCallerInput(stream VoiceVARequest) returns (stream VoiceVAResponse);

  /*
 The Service that takes virtual agent list request and org id and returns a list of bots.
 Returned list is shown on flow(virtual agent v2 activity) to select specific VA/Bot
*/
  rpc ListVirtualAgents(ListVARequest) returns (ListVAResponse) {}
}