SpeechSense dialogs
A dialog is a SpeechSense object that includes a recording of a call between a call center agent and a customer, as well as various metadata, such as the agent and customer full names, call date and language, etc. Calls are recorded with a phone exchange (PBX) and then uploaded to SpeechSense.
For each call, SpeechSense automatically separates the agent's speech and the customer's speech, applying different tags to their respective phrases. For instance, this may help to determine whether the agent greeted the customer or whether the latter was polite.
All dialogs break down into two categories:
- Outgoing: Those initiated by an agent.
- Incoming: Those initiated by a customer.
With SpeechSense, you can:
- View dialog specifications.
- Listen to the call recording.
- Understand how well an agent does their job.
- View the call transcript.
- See how the text is marked up with tags.
There are various ways to work with dialogs:
- Find a dialog and view it in the dialog list.
- Build a report based on dialogs.
Dialog filtering
Filters define the conditions for dialog search.
There are four types of filters:
- Operator: Defines how well the agent does their job, e.g., how fast they were speaking, whether they interrupted the customer, etc.
- Customer: Refers to the customer's behavior during the call, e.g., how fast they were speaking, whether they interrupted the agent, etc.
- General metadata: Recording data.
- Customer tags and Operator tags: Classifiers applied to the call text recognition results. You can learn more about tags here.
For each filter, you can specify one or more values, which are the filtering conditions. Such values fall into these three categories:
- Date: Select a date range from the calendar.
- Text: Enter some text. The search will only return exact matches.
- Number: Specify a range of numbers. You can specify either both range boundaries or just one of them. To find a certain value, specify it as both the higher and the lower boundary. The boundaries are included into the range being filtered.
You can use multiple filters at a time to find the dialogs that match all specified conditions.
Detailed info on dialogs
For each dialog, you have the following information available:
-
General info, e.g., who initiated the call and how long it lasted.
-
Information on the agent and customer.
-
Call recording.
-
Summary that YandexGPT API generates automatically based on full semantic analysis. Technically, it provides YandexGPT API's Yes/No answers to questions.
-
Recording transcript Yandex SpeechKit provides automatically.
Using a transcript, you can search through a text fragment in the agent's or customer's channel. The search will return exact matches. Found fragments are highlighted in yellow.
The transcript is automatically marked up with agent and customer tags.