metadata
dataset_info:
- config_name: bosnia-herzegovina
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: validation
num_bytes: 919337453.444
num_examples: 1909
- name: train
num_bytes: 77993681122.905
num_examples: 162735
- name: test
num_bytes: 943174268.733
num_examples: 1961
download_size: 79113448032
dataset_size: 79856192845.082
- config_name: bulgaria
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: 248065429445.564
num_examples: 513483
- name: validation
num_bytes: 3510710800.976
num_examples: 7294
- name: test
num_bytes: 3344762135.408
num_examples: 6892
download_size: 252824889652
dataset_size: 254920902381.948
- config_name: croatia
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: 316453728263.914
num_examples: 664881
- name: test
num_bytes: 2702349904.91
num_examples: 5635
- name: validation
num_bytes: 2699814032.448
num_examples: 5624
download_size: 322990100208
dataset_size: 321855892201.272
- config_name: denmark
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: 185531841060.95
num_examples: 398773
- name: validation
num_bytes: 2202312175.554
num_examples: 4646
- name: test
num_bytes: 1879996365.696
num_examples: 3981
download_size: 190728692038
dataset_size: 189614149602.2
- config_name: estonia
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: 101842775306.9
num_examples: 208412
- name: test
num_bytes: 1714742404.466
num_examples: 3554
- name: validation
num_bytes: 1528909434.7
num_examples: 3175
download_size: 103906641961
dataset_size: 105086427146.066
- config_name: finland
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: 32594336911.892
num_examples: 68639
download_size: 32855230508
dataset_size: 32594336911.892
- config_name: france
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: 9226444028.936
num_examples: 19296
- name: test
num_bytes: 352151108
num_examples: 744
download_size: 9491754427
dataset_size: 9578595136.936
- config_name: germany
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: 241755081139.9
num_examples: 502515
- name: test
num_bytes: 2355739115.28
num_examples: 4872
- name: validation
num_bytes: 2981301054.543
num_examples: 6193
download_size: 494097337342
dataset_size: 247092121309.723
- config_name: greece
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: 263056171312.21
num_examples: 562010
- name: validation
num_bytes: 4448185444.424
num_examples: 9262
- name: test
num_bytes: 3218188587.73
num_examples: 6730
download_size: 275818910894
dataset_size: 270722545344.364
- config_name: iceland
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: 71472785790.548
num_examples: 149533
- name: validation
num_bytes: 1310102178.86
num_examples: 2732
- name: test
num_bytes: 1507775016.649
num_examples: 3159
download_size: 77288074579
dataset_size: 74290662986.057
- config_name: italy
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: 313435161177.292
num_examples: 658661
- name: validation
num_bytes: 3574721188.877
num_examples: 7429
download_size: 317422730589
dataset_size: 317009882366.169
- config_name: latvia
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: 137776566199.313
num_examples: 285083
- name: validation
num_bytes: 1509002378.768
num_examples: 3188
- name: test
num_bytes: 1593198912.841
num_examples: 3343
download_size: 276643104304
dataset_size: 140878767490.922
- config_name: lithuania
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: 301683769737.85
num_examples: 631550
- name: validation
num_bytes: 3084006164.348
num_examples: 6434
- name: test
num_bytes: 3502221089.863
num_examples: 7319
download_size: 305733976991
dataset_size: 308269996992.061
- config_name: malta
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: 69838452620.337
num_examples: 147337
- name: validation
num_bytes: 588717209.4
num_examples: 1240
- name: test
num_bytes: 1065731467.56
num_examples: 2260
download_size: 70377616780
dataset_size: 71492901297.297
- config_name: norway
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: 440250112431.455
num_examples: 906667
- name: validation
num_bytes: 4856964807.244
num_examples: 10038
- name: test
num_bytes: 4861683338.004
num_examples: 10042
download_size: 441901533208
dataset_size: 449968760576.70306
- config_name: portugal
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: 368098713713.143
num_examples: 780469
- name: validation
num_bytes: 3744583303.471
num_examples: 7951
- name: test
num_bytes: 3548102274.206
num_examples: 7501
download_size: 379580153099
dataset_size: 375391399290.82
- config_name: serbia
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: 207935148745.124
num_examples: 432659
- name: validation
num_bytes: 2366199111.852
num_examples: 4853
- name: test
num_bytes: 2370938230.09
num_examples: 4961
download_size: 213167192459
dataset_size: 212672286087.06598
- config_name: slovakia
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: validation
num_bytes: 3176657949.876
num_examples: 6607
- name: test
num_bytes: 2469280736.248
num_examples: 5104
download_size: 5573778559
dataset_size: 5645938686.124001
- config_name: slovenia
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: 129438945198.308
num_examples: 270428
- name: validation
num_bytes: 1788987694.534
num_examples: 3721
- name: test
num_bytes: 1710926793.385
num_examples: 3585
download_size: 132999734304
dataset_size: 132938859686.22699
- config_name: sweden
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: 260176383382.708
num_examples: 543459
- name: validation
num_bytes: 3015372605.905
num_examples: 6305
- name: test
num_bytes: 2966220472.494
num_examples: 6138
download_size: 271304713141
dataset_size: 266157976461.107
- config_name: uk
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: 19587393098.801
num_examples: 40791
download_size: 19497150061
dataset_size: 19587393098.801
- config_name: ukraine
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: 269118203613.392
num_examples: 277032
- name: test
num_bytes: 3148902172.179
num_examples: 3239
- name: validation
num_bytes: 2977516808.2
num_examples: 3064
download_size: 248392546680
dataset_size: 275244622593.771
configs:
- config_name: bosnia-herzegovina
data_files:
- split: validation
path: bosnia-herzegovina/validation-*
- split: train
path: bosnia-herzegovina/train-*
- split: test
path: bosnia-herzegovina/test-*
- config_name: bulgaria
data_files:
- split: train
path: bulgaria/train-*
- split: validation
path: bulgaria/validation-*
- split: test
path: bulgaria/test-*
- config_name: croatia
data_files:
- split: test
path: croatia/test-*
- split: validation
path: croatia/validation-*
- split: train
path: croatia/train-*
- config_name: denmark
data_files:
- split: train
path: denmark/train-*
- split: validation
path: denmark/validation-*
- split: test
path: denmark/test-*
- config_name: estonia
data_files:
- split: train
path: estonia/train-*
- split: test
path: estonia/test-*
- split: validation
path: estonia/validation-*
- config_name: finland
data_files:
- split: train
path: finland/train-*
- config_name: france
data_files:
- split: train
path: france/train-*
- split: test
path: france/test-*
- config_name: germany
data_files:
- split: train
path: germany/train-*
- split: test
path: germany/test-*
- split: validation
path: germany/validation-*
- config_name: greece
data_files:
- split: train
path: greece/train-*
- split: validation
path: greece/validation-*
- split: test
path: greece/test-*
- config_name: iceland
data_files:
- split: validation
path: iceland/validation-*
- split: test
path: iceland/test-*
- split: train
path: iceland/train-*
- config_name: italy
data_files:
- split: train
path: italy/train-*
- split: validation
path: italy/validation-*
- config_name: latvia
data_files:
- split: train
path: latvia/train-*
- split: validation
path: latvia/validation-*
- split: test
path: latvia/test-*
- config_name: lithuania
data_files:
- split: train
path: lithuania/train-*
- split: validation
path: lithuania/validation-*
- split: test
path: lithuania/test-*
- config_name: malta
data_files:
- split: train
path: malta/train-*
- split: validation
path: malta/validation-*
- split: test
path: malta/test-*
- config_name: norway
data_files:
- split: train
path: norway/train-*
- split: validation
path: norway/validation-*
- split: test
path: norway/test-*
- config_name: portugal
data_files:
- split: train
path: portugal/train-*
- split: validation
path: portugal/validation-*
- split: test
path: portugal/test-*
- config_name: serbia
data_files:
- split: train
path: serbia/train-*
- split: validation
path: serbia/validation-*
- split: test
path: serbia/test-*
- config_name: slovakia
data_files:
- split: validation
path: slovakia/validation-*
- split: test
path: slovakia/test-*
- config_name: slovenia
data_files:
- split: train
path: slovenia/train-*
- split: validation
path: slovenia/validation-*
- split: test
path: slovenia/test-*
- config_name: sweden
data_files:
- split: validation
path: sweden/validation-*
- split: test
path: sweden/test-*
- split: train
path: sweden/train-*
- config_name: uk
data_files:
- split: train
path: uk/train-*
- config_name: ukraine
data_files:
- split: train
path: ukraine/train-*
- split: test
path: ukraine/test-*
- split: validation
path: ukraine/validation-*
task_categories:
- automatic-speech-recognition
- text-to-speech
language:
- de
- en
- mt
- sl
- sk
- da
- sv
- 'no'
- bs
- fr
- hr
- pt
- it
- lt
- el
- bg
- lv
- uk
- fi
- et
- is
- sr
EuroSpeech Dataset
Dataset Description
EuroSpeech is a large-scale multilingual speech corpus containing high-quality aligned parliamentary speech across 22 European languages. The dataset was constructed by processing parliamentary proceedings using a robust alignment pipeline that handles diverse audio formats and non-verbatim transcripts.
Dataset Summary
- Languages: 22 European languages (see detailed breakdown below)
- Total aligned hours: ~78,100 hours of initially aligned speech-text data
- Quality-filtered subsets:
- CER < 30%: approximately 61,000 hours
- CER < 20%: approximately 50,500 hours (this is the primary subset provided directly through the Hugging Face Datasets interface for all languages)
- CER < 10%: approximately 32,200 hours
- Domain: Parliamentary proceedings (formal speaking style)
- Audio segment length: Typically 3-20 seconds
- Format: Audio segments with paired transcriptions
Languages
EuroSpeech provides substantial data for previously under-resourced languages:
- 19 languages exceed 1,000 hours of data (CER < 20%)
- 22 languages exceed 500 hours of data (CER < 20%)
| Language | Code | Total Aligned (h) | CER < 30% (h) | CER < 20% (h) | CER < 10% (h) |
|---|---|---|---|---|---|
| Croatia | hr | 7484.9 | 5899.7 | 5615.8 | 4592.0 |
| Denmark | da | 7014.2 | 6435.0 | 5559.8 | 3443.7 |
| Norway | no | 5326.2 | 4578.8 | 3866.7 | 2252.2 |
| Portugal | pt | 5096.3 | 4036.7 | 3293.5 | 2105.9 |
| Italy | it | 4812.8 | 3539.6 | 2813.7 | 1767.3 |
| Lithuania | lt | 5537.9 | 3971.0 | 2681.2 | 956.6 |
| United Kingdom | en | 5212.2 | 3790.7 | 2609.3 | 1175.0 |
| Slovakia | sk | 2863.4 | 2722.4 | 2553.6 | 2070.8 |
| Greece | el | 3096.7 | 2717.6 | 2395.4 | 1620.9 |
| Sweden | sv | 3819.4 | 2862.6 | 2312.8 | 1360.1 |
| France | fr | 5476.8 | 2972.1 | 2249.8 | 1347.6 |
| Bulgaria | bg | 3419.6 | 2570.4 | 2200.1 | 1472.8 |
| Germany | de | 2472.2 | 2354.2 | 2184.4 | 1698.4 |
| Serbia | sr | 2263.1 | 1985.1 | 1855.7 | 1374.1 |
| Finland | fi | 2130.6 | 1991.4 | 1848.2 | 1442.2 |
| Latvia | lv | 2047.4 | 1627.9 | 1218.8 | 499.9 |
| Ukraine | uk | 1287.8 | 1238.3 | 1191.1 | 1029.8 |
| Slovenia | sl | 1338.2 | 1241.7 | 1156.4 | 900.5 |
| Estonia | et | 1701.1 | 1430.9 | 1014.9 | 382.5 |
| Bosnia & Herz. | bs | 860.2 | 781.9 | 691.3 | 447.8 |
| Iceland | is | 1586.1 | 974.1 | 647.4 | 171.4 |
| Malta | mt | 3281.6 | 1284.3 | 613.0 | 143.9 |
| Total | 78128.6 | 61006.4 | 50572.9 | 32255.5 |
Dataset Structure
Data Instances
Each instance in the dataset consists of:
- Audio segment (3-20 seconds)
- Corresponding transcript text
- Metadata including language, source session, alignment quality metrics
Data Splits
The dataset provides predefined train, development, and test splits for each language. To ensure data integrity and prevent leakage between sets, these splits are constructed by assigning entire parliamentary sessions (i.e., all segments derived from a single original long audio recording) exclusively to one of the train, development, or test sets. The exact proportions follow common practices (e.g., 80/10/10).
Dataset Creation
Source Data
The data was collected from parliamentary proceedings across 22 European nations. Parliamentary sessions offer high-quality speech in a formal register, typically featuring clear speech with good audio quality and professional transcripts.
Data Collection and Processing
The dataset was constructed using a multi-stage pipeline:
Data Sourcing and Metadata Collection: Manual and scripted gathering of media/transcript links from parliamentary websites.
Download Pipeline: Automated retrieval of audio, video, and transcript files using specialized handlers for diverse source formats.
Alignment Pipeline:
- Segmentation of long recordings into 3-20 second utterances using voice activity detection (VAD)
- Transcription of segments using an ASR model to produce pseudo-labels
- Alignment of segments to transcripts using a novel two-stage dynamic algorithm
- Selection of best-aligned transcript formats and quality filtering
Filtering: CER-based filtering to create quality tiers (CER < 30%, < 20%, < 10%)
Alignment Algorithm
The core of the alignment process is a novel two-stage dynamic algorithm specifically engineered for extreme robustness when matching ASR pseudo-labels to noisy, non-verbatim parliamentary transcripts:
Coarse stage: Uses a sliding window to rapidly scan the transcript, efficiently bypassing large irrelevant sections to identify a set of top-k candidate text spans via Character Error Rate (CER).
Fine-tuning stage: Performs a local search around promising candidates, optimizing start position and window size for the best CER.
A fallback mechanism restarts the search if no initial match meets a predefined quality threshold.
Dataset Use
Intended Uses
The EuroSpeech dataset is intended for:
- Training and evaluating automatic speech recognition (ASR) systems
- Training and evaluating text-to-speech (TTS) systems
- Multilingual speech research
- Low-resource language speech technology development
- Cross-lingual transfer learning in speech models
Citation Information
If you use this dataset, please cite:
[Citation details to be added upon publication]
Considerations
Data Quality
The dataset provides multiple quality tiers based on Character Error Rate (CER):
- CER < 30%: More data, but potentially lower quality alignments
- CER < 20%: Balanced quality-quantity trade-off (recommended for most applications)
- CER < 10%: Highest quality alignments, but reduced quantity
Licensing Information
[License details to be added]
Limitations
- The dataset primarily represents formal parliamentary speech and may not generalize well to casual, spontaneous, or noisy speech environments.
- The dataset reflects the demographics and speaking styles of European parliamentarians, which may not be representative of the general population.
- Some languages have significantly more data than others, which could lead to performance disparities in multilingual models.
Additional Information
Dataset Curators
- Samuel Pfisterer (@SamuelPfisterer1)
- Florian Grötschla (@FloGr)
- Luca Lanzendörfer (@lucala)
- Florian Yan (@floyan)
- Roger Wattenhofer
Maintenance Status
[Information about maintenance and update plans]