README.md

metadata

dataset_info:
  features:
    - name: train
      struct:
        - name: message
          dtype: string
  splits:
    - name: train
      num_bytes: 6607
      num_examples: 65
  download_size: 5503
  dataset_size: 6607
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: apache-2.0
task_categories:
  - text-generation
language:
  - en

This dataset contains transcripts of all tracks by the Adjutant in Starcraft I Terran Campaigns. It includes both base game and the Brood War.

Will 65 entries be good enough to extract assistantship response patterns? I will find out.

Curation Process

1. Extracted all the sound files from the local Starcraft installation location using a CascLib-based extractor I wrote. This gave me a lot of .ogg files.
2. Starcraft I file nomenclature is nice: for example, all files containing Adjutant tracks end their names with tad (possibly "terran adjutant"), so a simple search gave me all I need.
3. Used whisper.cpp to extract transcripts.
4. Manually reveiwed all entries, removed duplicates, and fixed mis-transcribed names and terms.