Skip to content

Voice mode / audio transcription support #6

Description

@marklubin

ChatGPT voice mode conversations store message content as dict parts (content_type: "audio_transcription" with a "text" field, plus audio_asset_pointer and real_time_user_audio_video_asset_pointer types) instead of plain strings. The adapter currently only extracts isinstance(p, str) parts, so voice conversations are silently dropped (0 text parts → skipped).

Fix: extract .text from audio_transcription dict parts. This recovers 2 of 10 conversations in the current test export.

Metadata

Metadata

Assignees

No one assigned

    Labels

    adaptersAdapters & parsingbacklogDeferred from v0.9enhancementNew feature or request

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions