Data flow
V‑Spark is highly configurable for a wide variety of uses cases. The following generalized workflow does not cover every use case, but it illustrates how audio data moves through the system.
- Audio and metadata file preparation.
- Ensure that audio is compatible with V‑Spark's built-in audio evaluation tool. V‑Spark supports one- or two-channel audio only; audio with more than two channels will be rejected.
Ensure metadata is correctly formatted and does not violate any system metadata restrictions.
- Audio and metadata upload.
-
Audio may be uploaded with the UI or with the API.
Custom metadata may be uploaded with audio data in a zip, or after the audio data in a separate upload.
-
- V‑Spark queues the audio for transcription.
- ASR transcription.
-
ASR settings are configured for each folder.
Available options depend on the folder's company and organization configuration and licensing.
-
- Transcription output queued for analysis by V‑Spark.
- JSON transcript analytics processing.
-
Basic metrics provided for all transcripts.
-
Custom analysis performed by applications depending on configuration.
-
- Results available for download.
- If callbacks are enabled, V‑Spark sends results to the configured destination.
Other important workflows:
- Reprocessing audio
- Re-scoring transcripts
- Deleting audio, transcription, and analytics data
- Scheduled and ad-hoc report generation