Skip to content

[asr] feature(asr-worker): unify ASR workflow output and save transcriptions as artifacts #9

@ClemDoum

Description

@ClemDoum

Currently ASRActivities.infer return ParakeetModelHandlerResult, these should be more generic to support output of all ASR implementations, it should contain:

  • text with timestamps
  • diarization info (optional)

Additionally the workflow should persist these in as artefact. Potentially the JSONL format is a good format to be able to read results without deserializing the full transcript for long files

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

Status

In Progress

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions