README.md

---
dataset_info:
  config_name: latvia_0.06_cer_filtered
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
  - name: key
    dtype: string
  - name: country
    dtype: string
  - name: language
    dtype: string
  - name: video_id
    dtype: string
  - name: transcript_id
    dtype: string
  - name: start_seconds
    dtype: float32
  - name: end_seconds
    dtype: float32
  - name: duration_seconds
    dtype: float32
  - name: asr_transcript
    dtype: string
  - name: human_transcript
    dtype: string
  - name: cer
    dtype: float32
  - name: wer
    dtype: float32
  - name: original_transcript_start_idx
    dtype: int32
  - name: original_transcript_end_idx
    dtype: int32
  splits:
  - name: train
    num_bytes: 22952837865.97
    num_examples: 48930
  - name: validation
    num_bytes: 184914678.0
    num_examples: 401
  - name: test
    num_bytes: 200823453.0
    num_examples: 424
  download_size: 23005841186
  dataset_size: 23338575996.97
configs:
- config_name: latvia_0.06_cer_filtered
  data_files:
  - split: train
    path: latvia_0.06_cer_filtered/train-*
  - split: validation
    path: latvia_0.06_cer_filtered/validation-*
  - split: test
    path: latvia_0.06_cer_filtered/test-*
---