metadata
dataset_info:
features:
- name: start
dtype: int64
- name: end
dtype: int64
- name: duration
dtype: int64
- name: text
dtype: string
- name: text_normalized
dtype: string
- name: text_whisper
dtype: string
- name: whisper_transcription
dtype: string
- name: wav2vec_transcription
dtype: string
- name: bleu_whisper
dtype: float64
- name: bleu_wav2vec
dtype: float64
- name: wer_whisper
dtype: float64
- name: wer_wav2vec
dtype: float64
- name: whisper_first
dtype: float64
- name: wav2vec_first
dtype: float64
- name: whisper_last
dtype: float64
- name: wav2vec_last
dtype: float64
- name: speech_id
dtype: string
- name: protocol_id
dtype: string
- name: sub_ids
sequence: int64
- name: chunk_id
dtype: int64
- name: lang_prob_sv
dtype: float64
- name: audio_file
dtype: string
- name: name
dtype: string
- name: party
dtype: string
- name: gender
dtype: string
- name: role
dtype: string
- name: district
dtype: string
- name: speaker_id
dtype: string
- name: riksdagen_id
dtype: string
- name: dates
sequence: string
- name: date_approx
dtype: timestamp[ns]
- name: year
dtype: int64
- name: shard
dtype: string
- name: audio
dtype:
audio:
sampling_rate: 16000
splits:
- name: train
num_bytes: 78075901
num_examples: 100
download_size: 77991004
dataset_size: 78075901
configs:
- config_name: default
data_files:
- split: train
path: data/train-*