V‑Spark applications enable robust analysis of ASR transcribed audio interactions. Use applications to

  • Classify audio by a particular application or category score, or by metadata value.
  • Find and further analyze particular types of calls like service cancellations or renewals.
  • Score and organize audio by call center performance metrics and other considerations, like first call resolution, agent performance, policy compliance, call driver analysis.

V‑Spark applications are defined by categories, phrases, and filters that examine transcribed speaker phrases or audio metadata values.

V‑Spark uses applications to analyze transcripts in terms of specific speech phrases and conversation patterns. This analysis is highly configurable.

Data points from transcript-application analysis are called scores. There are two types of score: coverage, and hit-or-miss.

Applications are linked to folders. Folders receive and transcribe audio. A folder analyzes all of its transcription data with every linked application.

In addition to analyzing incoming audio, applications can reprocess and rescore all of the audio in a folder. If an application is modified, application scores do not change until the application reprocesses the transcripts.

V‑Spark application scores may be aggregated and reviewed with daily and monthly charts in the V‑Spark UI, or reviewed in other analysis tools with exported CSV reports.

There are several ways to work with V‑Spark applications and scores:

  • Create, delete, or change the settings for an application with the Settings page.
  • Create and manage categories and phrase filters with the Application Editor.
  • Download and upload application definition JSON files with the UI or API.
  • Link folders and applications to automatically process audio sent to a folder.
  • Reprocess audio to obtain new application scores.