Skip to content

Voicenote Storage: Optimize audio handling #804

@nishika26

Description

@nishika26

Describe the current behavior
Voicenote audio is stored as base64-encoded strings directly in the database — for STT evaluation and LLM/LLM chain calls. Each voicenote can be several MBs, meaning every row carrying audio bloats the database significantly and makes queries slower.

Describe the enhancement you'd like
Upload voicenote audio to S3 at ingestion time and store only the S3 URI in the database, following the same pattern already used for documents. The base64 payload should never be persisted to the database.

Metadata

Metadata

Assignees

Labels

enhancementNew feature or request

Type

Projects

Status

In Progress

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions