Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html> <h"... is not valid JSON

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

fdt-disfluency-synthetic

Synthetic disfluency-removal data: clean conversational text (DailyDialog + Synthetic-Persona-Chat) with disfluencies injected programmatically, giving exact word-level labels (KEEP / DELETE / KEEP_STRIP_COMMA / KEEP_CAPITALIZE). Invariant: applying the tags reconstructs the clean text exactly. Rows: {{"words": [...], "tags": [...], "clean": "..."}}.

  • v1/ — fillers (um, uh, er, ah, like, you know, I mean, sort of, kind of, basically, literally), repetitions, false starts, stutters/cut-offs, lowercase-STT variants, adversarial keep-templates.
  • v2/ — v1 plus substitution repairs ("the bus, I mean, the train"), filler-homograph collisions ("Do you, like, like jazz?"), doubled fillers, denser disfluency stacks.
  • eval/adversarial.jsonl — 65 handwritten cases incl. out-of-pool quarantine classes for measuring schema transfer.

License note: DailyDialog is CC BY-NC-SA, so this dataset inherits non-commercial terms. Generator code lives in the FluencyAI repo (ml/disfluency/disfl/injector.py).

Downloads last month
29