Sentiment by Turn

Overview

Sentiment by Turn allows the user to track sentiment between the caller and agent on stereo calls. For more value, Sentiment by Turn may be configured with Verb-Noun Pairs so the user can understand the topical drivers of positive and negative sentiment.

Prerequisites for Use

  • Calls MUST be processed in stereo for agent and caller specific metrics (otherwise, these metrics always return 0)
  • The Agent SHOULD have a speaker name that is one of: Agent, Service, Representative, Operator, Salesperson, CallCenter, ContactCenter
  • The Caller SHOULD have a speaker name that is one of: Caller, Client, Customer, Prospect
  • If the speaker names are not specified as above, the first speaker is assumed to be the agent, and the second speaker is assumed to be the caller
  • Advanced Punctuation MUST be enabled.
  • English and Spanish are the only supported languages for this feature.

Configuration

{
  "speechModel": {
    "language": "es-US",
    "features": [  "advancedPunctuation"  ]
  },
  "ingest": {
    "stereo": {
       "left": { "speakerName": "Caller" },
      "right": { "speakerName": "Agent"    }
    }
  },
   
   "speakerSentiments":true

}

Output

{ 
  "speakerSentiments": [
    {
      "speakerName": "Caller",
      "sentimentValues": [
        {
          "s": 4558,
          "e": 7064,
          "v": -0.5434
        }, {
          "s": 9373,
          "e": 10345,
          "v": 0.7039
        }
      ]
    }, {
      "speakerName": "Agent",
      "sentimentValues": [
        {
          "s": 7464,
          "e": 9373,
          "v": 0.4328
        }, {
          "s": 12937,
          "e": 14627,
          "v": -0.3294
        }
      ]
    }
  ]
}

The sample output above contains the fields s,e,and v, which are outputs from the VoiceBase sentiment engine. Their definitions are as follows:

  • s: The start timestamp of the segment in milliseconds
  • e: The end timestamp of the segment in milliseconds
  • v: The sentiment score assigned to the segment

For more information on sentiment values, see Conversation Metrics.