Meta: End-to-End Audio Processing Pipelines

[satra]

Umbrella issue for complete audio processing workflows.

Sub-issues

• Task: Add workflow for full audio -> diarized transcript pipeline #350 Full audio → diarized transcript pipeline (needs forced alignment integration)
• Task: Improve multi-speaker diarization consistency across audio segments #378 Improve multi-speaker diarization consistency across audio segments
• Task: Improve background noise handling for child speech recordings #377 Improve background noise handling for child speech recordings

Scope

Build robust end-to-end pipelines: complete the diarized transcript workflow by integrating forced alignment (#350), improve multi-speaker consistency (#378), and handle challenging recording conditions like child speech with background noise (#377).

Current state

• explore_conversation workflow exists with diarization + transcription + features
• Forced alignment module exists but isn't integrated into the pipeline
• Speech enhancement exists but needs tuning for child speech scenarios