Is spoken part clearly identifiable? No - no genre-like field or document_id found at all; no comment-level metadata beyond sent_id/text detected. Flagged to maintainers to confirm which sentences/documents are spoken vs. written.
Metadata review
languages and translation(s)
Field
Advice
text_en
OK (ISO 639-1 two-letter code)
transcription and annotation levels available
Field
Advice
text_ortho
change to text_orthographic
speaker metadata
Field
Advice
speaker_id
OK
sent metadata
Field
Advice
sent_timecode
split into sound_alignment_begin, sound_alignment_end, and duration