Skip to content

ADR-v2-091 — Spoken Table → CSV / Field Data Entry

Status: Accepted (2026-07-02) · Wave K Context links: [[adr-v2-059-spoken-spreadsheet]] (nav/formulas vs bulk entry), [[adr-v2-063-slot-filling-dictation]], [[adr-011]]

Context

Wave K research (#7) — a row/column dictation mode: "row: Ada, 1815, London" injects tab/comma- separated cells and advances; "next row" moves down. For spreadsheets and forms. Distinct from Spoken Spreadsheet (cell navigation/formulas) — this is bulk structured row/field data entry with a delimiter + Tab/Enter cadence, a data-entry accessibility workflow. Anchor: Vocal Forms (IJSREM 2025), VaaniSevak offline Vosk census entry.

Decision

Add an opt-in Spoken Table→CSV: [tablecsv] enabled=false, delimiter=",". Pure cores: parse_row(text) strips a leading "row/entry/record" marker and splits into cells; rows_to_delimited(text, sep) splits multiple "next row"-separated rows into delimited lines; and row_plan(cells, cell_key, row_end) emits an injection plan (type cell → press Tab, … → press Enter). Dependency-free. OFF by default.

Consequences

  • Bulk hands-free row/field entry for spreadsheets and forms; pure parse + cadence.
  • Distinct from Spoken Spreadsheet (bulk entry vs navigation).
  • Privacy (ADR-011): local text + keystrokes only.
  • Caveat: cell splitting is on commas/"and" → free-form cell text with commas needs quoting (a later tier); off by default.