Skip to content
INDXR.AI
All docs

CSV

The CSV export turns your transcript into a spreadsheet — one row per segment — so you can sort it, count words, filter by time, or load it into Excel, Google Sheets, or pandas. Each row carries the segment's timing and text; the exact columns are segment_index, start_time, end_time, duration, word_count and text, with a UTF-8 byte-order mark so Excel opens it correctly.

Columns

Scroll the table horizontally →

ColumnMeaning
segment_indexPosition of the segment in the transcript, counted from 0 (the first segment is 0)
start_timeSegment start, in seconds
end_timeStart of the next segment (or clip end for the last)
durationSegment duration, in seconds
word_countWords in the segment text
textThe segment text (quoted; internal quotes doubled)

Output

The # lines at the top carry the video's metadata; they are comments, not data rows. Video metadata is not repeated per row.

# title: Vector Databases Explained
# url: https://www.youtube.com/watch?v=dQw4w9WgXcQ
# channel: AI Foundations
# published: 2026-06-01
# duration_seconds: 31
# language: en
# transcript_source: AI Transcription (AssemblyAI)
# extracted: 2026-07-22
segment_index,start_time,end_time,duration,word_count,text
0,0,3.4,3.4,8,"Welcome to this short introduction to vector databases."
1,3.4,7.5,4.1,10,"A vector database stores embeddings and retrieves them by similarity."
2,7.5,12.1,4.6,11,"That makes it the backbone of most retrieval augmented generation systems."

Sources

  • INDXR (own code) — column names, metadata comment lines, BOMVerified against packages/shared/src/utils/formatTranscript.ts (generateCsv)

See also