Hugging Face's logo

Datasets:
changelinglab
/
speechocean-l2eval

Tasks:
Automatic Speech Recognition
Modalities:
Audio
Text
Formats:
parquet
Languages:
English
Size:
1K - 10K
ArXiv:
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
audio
audioduration (s)
1.59
15.3
speaker_id
stringclasses
113 values
utt_id
stringlengths
5
8
text
stringlengths
3
60
accuracy
int32
3
10
completeness
float32
6.7
10
fluency
int32
1
10
prosodic
int32
1
10
total
int32
2
10
1
10011
WE CALL IT BEAR
8
10
9
9
8
1
10035
ZERO THREE FIVE ONE
8
10
9
9
8
1
10053
THREE TWO TWO SEVEN
9
10
10
10
9
1
10063
ELEPHANTS TAI GOOSE
10
10
9
9
9
1
10069
TOM GIVES UP BOXING
9
10
9
10
9
1
10075
HE HATES SHOOTING
9
10
9
9
9
1
10089
MANDY HAS A BIG ARM
10
10
9
9
9
1
10095
LOOK AT ANN'S PANTS
10
10
9
9
9
1
10106
WHAT ABOUT THE BUS
9
10
9
8
8
1
10113
THEN HE WENT TO THEME PARK
6
10
9
8
5
1
10115
LET'S GO TO THE RESTROOM
9
10
9
9
9
1
10121
THEN MIKE WALKS TO COFFEE
9
10
9
9
8
1
10122
SO MARY WENT ON TO STUDY
9
10
9
9
9
1
10133
KATE GOT THE TOMATO
8
10
8
8
8
1
10135
TINA LOVES EGGPLANT
8
10
9
9
8
1
10140
DORA IS NOT A CLEANER
8
10
9
9
8
1
10145
MARK LIVED IN NEW YORK
9
10
9
9
9
1
10149
BOB LIVES IN CAIRO NOW
8
10
9
8
8
1
10168
BYE
10
10
10
10
10
1
10173
TREES
9
10
10
9
9
5
50003
MIKE LIKES THE WHITE ONE
8
10
6
7
7
5
50010
ITS NAME IS SAY
7
10
7
7
7
5
50024
BILLY LOVES AMERICA
7
10
6
6
6
5
50028
TWO TWO EIGHT SEVEN
8
10
8
8
8
5
50038
FOUR FIVE FOUR SEVEN
7
10
8
7
7
5
50040
FIVE THREE NINE ZERO
8
10
8
7
8
5
50047
SIX FIVE THREE
7
10
8
7
7
5
50049
TWO FIVE EIGHT
7
10
9
8
7
5
50055
WE LESS MEAT
9
10
8
8
8
5
50078
DOSE MIKE LIKE THE HAMBURGER
7
10
7
6
7
5
50079
ANNIE WANT HAVE SOME PRETTIES
7
10
6
6
6
5
50083
KATE'S GOT SOME GREY
6
7.5
7
7
5
5
50095
LOOK AT JOHN'S SWEATER
7
10
6
7
7
5
50099
TIM HAS A BEAUTIFUL TELL
7
10
8
6
7
5
50100
JIM LIKES YOUR BLUE SHORE
8
10
8
7
7
5
50114
ANNIE WENT WALKING THEM PARK
8
10
8
6
7
5
50118
LAYLA CAN SEE THE HOMETOWN
6
10
7
6
5
5
50122
SO TINA WENT ON TO WASHROOM
6
10
6
5
5
5
50174
ALL WITH HIM
8
10
7
6
7
5
50175
GOOD JOB
9
10
9
7
8
6
60015
JAYME IS GOING TO SEE HEN
9
10
9
9
9
6
60020
DORA IS COME FROM JAPAN
9
10
9
8
8
6
60029
SIX ONE SIX ZERO
9
10
9
9
9
6
60031
FIVE TWO FOUR SEVEN
9
10
9
9
9
6
60049
ZERO THREE FIVE
8
10
9
9
8
6
60056
PHONE PARTS TICKETS
10
10
9
9
9
6
60077
ANN ATE A LITTLE DOG
9
10
9
9
9
6
60081
DOES LAYLA LIKE THE JAM
10
10
10
9
9
6
60082
KATE LIKES LAMB
9
10
9
9
9
6
60083
MARK GOT SOME NOODLES
10
10
9
9
9
6
60094
TOM LIKES THE OLD SWEATER
10
10
9
9
9
6
60102
BOB NEEDS NEW BOOTS
10
10
10
9
9
6
60106
TOM CAN SEE THE MOTOR CYCLE
10
10
10
9
9
6
60111
HE LEFT THE FRUIT STAND
10
10
9
8
9
6
60113
LET'S GO TO THE GYM
10
10
9
8
9
6
60116
SO MARK WHEN TO THE PET SHOP
10
10
10
9
9
6
60124
SO ALICE WENT ON TO VILLAGE
9
10
8
9
9
6
60130
NO THAT'S TOMATO
10
10
9
9
9
6
60136
I'M NOT A ACTOR
9
10
9
8
9
6
60153
WHAT LOVELY CLEAN TEETH
9
10
9
8
9
26
260001
LAYLA LOVE BROWN
6
10
8
8
6
26
260011
ANDY CAN SEE THE HEN
8
10
8
9
8
26
260015
JOHN IS GOING TO SEE COW
9
10
9
9
9
26
260032
ONE ONE ZERO EIGHT
8
10
9
9
8
26
260033
THREE ONE SEVEN NINE
9
10
9
9
8
26
260039
EIGHT ZERO SIX TWO
9
10
9
9
9
26
260048
THREE ZERO EIGHT SEVEN
9
10
9
9
8
26
260050
ZERO EIGHT THREE ONE
9
10
9
9
9
26
260052
SIX EIGHT SIX SEVEN
10
10
9
9
9
26
260069
TEDDY GIVES UP SHOOTING
7
10
8
8
7
26
260075
HE LIKES RACING
9
10
9
8
8
26
260091
DORA LIKES YOUR RED JEANS
8
10
9
8
8
26
260095
BOB LOVES THE NEW CLOTH
8
10
9
9
8
26
260096
JOHN NEEDS NEW T SHIRT
9
10
9
9
9
26
260112
THEN LILLY WALKS TO MUSIC ROOM
9
10
9
9
9
26
260115
THEN TIM WALKS TO THEM PARK
9
10
8
8
8
26
260121
THEN MANDY WALKS TO RESTROOM
9
10
9
9
9
26
260126
DO YOU WANT THE PEA
9
10
9
9
9
26
260133
ANDY GOT THE MUSHROOM
10
10
9
9
9
26
260166
NOW FOR THE PARTY
9
10
9
9
9
36
360013
IT'S JUST SO HARD TO PICTURE
9
10
9
9
9
36
360034
WE SPEAK OUT WHEN WE FEEL WE SHOULD SPEAK OUT
9
10
9
8
8
36
360036
I COULD DO WITH A BREAK
9
10
9
8
8
36
360132
IT IS ALWAYS SO NICE TO SEE YOU
9
10
9
8
8
36
360133
UNFORTUNATELY NO ONE IS ALLOW TO SAY
8
10
9
8
8
36
360161
ALMOST TIME FOR A HAIRCUT
7
10
9
8
7
36
360190
SO YOU WANT TO BE MORE PRODUCTIVE
9
10
8
7
8
36
360210
STARTING CAN NOT BE BETTER THAN THIS
9
10
9
8
8
36
360223
I KEEP IT TO BE AMUSED BY THE STUPIDITY
7
10
8
8
7
36
360241
AFTER THAT I BE READY TO GO
8
10
8
8
8
36
360264
I THANK YOU FOR YOUR SYMPATHY BECAUSE IT COUNTS
7
10
8
8
7
36
360283
OFF THE CHAIN I BET
8
10
9
8
8
36
360313
HE NODDED HIS HEAD AND SMILED
8
10
9
8
8
36
360314
BUT TELL ME THE TRUTH
9
10
9
8
8
36
360332
THE FIRST QUESTION WAS AN OBVIOUS ONE
9
10
9
8
8
36
360334
THE SHOCK WAS TOO MUCH FOR HIM
9
10
9
8
8
36
360339
HE KNOWS HE'S ABOUT TO GET INTO FIGHT
8
10
9
8
8
36
360343
HE WAS ALWAYS READY TO DO ANYTHING WHATEVER
9
10
9
8
8
36
360360
THE LONG AND THE SHORT OF IT IS THIS
7
10
9
8
7
36
360378
YOU WANT TO BE LOVE
9
10
9
8
8
End of preview. Expand in Data Studio

speechocean762: A non-native English corpus for pronunciation scoring task

Dataset Summary

speechocean762 is an open-source non-native English speech corpus designed for pronunciation assessment and L2 spoken proficiency modeling. This Hugging Face version provides sentence-level audio and expert scores, organized into standard train / validation / test splits.

All speakers are Mandarin L1 learners of English, spanning both children and adults. Each utterance is evaluated independently by five expert annotators using standardized pronunciation metrics.

This dataset is suitable for:

  • pronunciation scoring
  • L2 speech assessment
  • speech representation learning
  • downstream regression or classification tasks

Dataset Structure

Splits

The dataset is published with three predefined splits:

  • train (2260)
  • val (240)
  • test (2500)

Splits are speaker-disjoint and provided as native Hugging Face splits.

Features

Each example contains:

Field Type Description
audio Audio Speech waveform (16 kHz)
speaker_id string Speaker identifier
utt_id string Utterance identifier
text string Prompt sentence
accuracy int Sentence-level pronunciation accuracy
completeness float Percentage of correctly pronounced words
fluency int Sentence-level fluency score
prosodic int Sentence-level prosody score
total int Overall pronunciation score

Scoring Metrics (Sentence level)

All sentence-level scores follow the original speechocean762 definitions. For detailed descriptions, see:

Dataset Creation

This Hugging Face dataset is derived from the original speechocean762 corpus and includes:

  • sentence-level audio
  • sentence-level expert scores
  • standardized HF Audio features
  • speaker-disjoint train/val/test splits

Word-level and phoneme-level annotations are not included in this version.

Source Dataset: https://huggingface.co/datasets/mispeech/speechocean762

License

The original speechocean762 dataset is released for free use, including commercial and non-commercial purposes, as stated by the original authors. Users should consult the original repository for full licensing details.

Citation

If you use this dataset, please cite the original paper:

@inproceedings{zhang2021speechocean762,
  title={speechocean762: An Open-Source Non-native English Speech Corpus For Pronunciation Assessment},
  author={Zhang, Junbo and Zhang, Zhiwen and Wang, Yongqing and Yan, Zhiyong and Song, Qiong and Huang, Yukai and Li, Ke and Povey, Daniel and Wang, Yujun},
  booktitle={Proc. Interspeech 2021},
  year={2021}
}

Acknowledgements

All credit for data collection and annotation belongs to the original speechocean762 authors. This Hugging Face release focuses on standardized access and reproducibility for modern speech and representation learning pipelines.

Downloads last month
17
Size of downloaded dataset files:
610 MB
Size of the auto-converted Parquet files:
610 MB
Number of rows:
5,000

Collection including changelinglab/speechocean-l2eval

Paper for changelinglab/speechocean-l2eval