README.md

---
dataset_info:
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
  - name: sentence
    dtype: string
  - name: prev_sentence
    dtype: string
  - name: episode_id
    dtype: string
  - name: start_time
    dtype: float64
  - name: end_time
    dtype: float64
  - name: token_count
    dtype: int32
  - name: quality_nll_loss
    dtype: float64
  - name: quality_nll_score
    dtype: float64
  - name: quality_audio_signature
    dtype: int64
  - name: quality_duplicate_count
    dtype: int64
  - name: quality_duplicate_score
    dtype: float64
  - name: quality_repetition_score
    dtype: float64
  - name: quality_diversity_score
    dtype: float64
  - name: quality_text_clean_score
    dtype: float64
  - name: quality_text_hebrew_ratio
    dtype: float64
  - name: quality_text_punct_ratio
    dtype: float64
  - name: quality_text_max_char_run
    dtype: int64
  - name: quality_audio_exact_dup_count
    dtype: int64
  - name: quality_audio_exact_dup_score
    dtype: float64
  - name: quality_audio_near_dup_count
    dtype: int64
  - name: quality_audio_near_dup_score
    dtype: float64
  - name: quality_audio_duplicate_score
    dtype: float64
  - name: quality_audio_contained_count
    dtype: int64
  - name: quality_audio_contained_ratio
    dtype: float64
  - name: quality_audio_containment_score
    dtype: float64
  - name: quality_token_rate
    dtype: float64
  - name: quality_token_rate_score
    dtype: float64
  - name: quality_aggregate_score
    dtype: float64
  splits:
  - name: train
    num_bytes: 147843090066
    num_examples: 287773
  download_size: 147721997960
  dataset_size: 147843090066
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---