Datasets:

ming030890
/

mdcc

sex stringclasses 2 values	duration float64 0.22 15	transcript stringlengths 1 80	id int64 0 25.6k
male	5.04	誘惑比佢稍小嘅人類始祖亞當同埋夏娃	0
male	1.05	王耀祖	1
male	3.14	佢從前食晚飯嘅時候總係喜歡飲啲酒	2
male	2.5	發現原本舖好嘅草皮	3
female	7.90001	好啊溫故而知新只要唔係懷舊回憶總係有好處嘅	4
female	5.38	我似乎到八九歲後才知道還有水桶這種器皿	5
male	1.64	即刻將我撞跌咗	6
female	13.07	只係最極端而已而呢一個年輕人因為呢個浪漫豪舉而囊空如洗當晚就冇錢食飯喇半工讀有可能嗎想自費留學而錢唔夠	7
female	3.11	一個鄉民嘴角担住一根香煙	8
female	3.22	騎馬嘅話一頓飯嘅功夫就可以到	9
male	3.92	呢一個係美學屈從於宗教法則嘅時代	10
female	1.42	去其他地方啦	11
female	3.16	每個人都將書擺喺桌嘅中央	12
male	0.57	佢	13
male	0.8	有一次	14
female	3.06	啊白抽紗嘅襯衣上裝	15
female	1.98	但係佢唔覺得懊喪	16
female	3.7	寫回歸女人一族嘅細碎事	17
male	4.1	入嚟之後就坐咗喺收銀檯前面嘅桌椅	18
female	4.12005	對於晚餐舞會{隨妻}更冇參加	19
female	1.44	又轉下食第二樣嘢	20
male	1.43	有聲好書	21
female	1.9	大家都好驚小卉知道	22
female	1.97	媽媽嘅面色有啲蒼白	23
female	5.06	正是我在朋友家晚餐席上遇到的那位小商人	24
female	2.24	只係偶然街上巧遇	25
female	4	日本軍國主義政府打擊共產主義思想	26
female	2.62	睇睇佢嘅衣裳靚定係唔靚	27
male	3.51	係乜嘢令佢哋由人變成為野獸	28
male	1.7	如果我回來	29
male	5.95	此後每日有一定數目嘅內地居民嚟香港同家人團聚	30
male	1.15	只重耐力	31
male	1.41	有聲好書	32
male	1.6	向晚悲風起	33
female	4	淡淡醬汁與純白豆腐互不奪色	34
male	1.96	我提到喺七十年代間	35
male	2.55	淡藍嘅天乾乾淨淨咁	36
male	5.14	嗰度嘅乾絲比起南京嘅細得多又從來冇咁甜	37
male	2.46	聲音演繹方梓豪	38
female	5.67001	那孩子剛一咧嘴笑那笑得才難看呢	39
female	2.34	班長幫我責罵陳小英	40
male	0.86	對唔住	41
male	4.1	佢哋既令人佩服但係既可憐又可笑	42
female	1.9	咁唔係一個簡單嘅民族	43
female	1.5	如電光一閃	44
female	3.90002	我都覺佢扮大個裝成熟	45
male	3.68	若果唔跟大佬自己根本冇法生存	46
female	5.2	喺呢個小鍋裏面打都唔知道比用牙咬快幾多倍喇	47
male	5.05	佢唔提對佢有恩而權力亦都係一度最大嘅呂不韋	48
female	1.89	我摸到一個小玻璃樽	49
male	2.66	足以改變一個人嘅信念	50
female	1.76	小蘭對住細佬妹話	51
female	2.23	去柏林查大學嘅記錄	52
male	1.22	包括霸主	53
male	3.56	如今卻係靠自己嘅本領賺返嚟嘅	54
male	1.37	{普連波}忍唔住笑	55
female	4.05	呢一個真係睜著眼睛說瞎話	56
male	5.02	所以我哋返屋企嘅時候魚仲係鮮蹦活跳㗎	57
female	4.7	臨瞓覺嘅時候指住爸爸小聲同母親話	58
female	1.84	淚水又流出嚟喇	59
female	7.22999	我根據西方觀念衡量真不曉得為什麼他們一直不開飯	60
female	3.84	且店外有一小牌寫住可按鈴叫人	61
female	4.72	祖父於是又換一個換一個唔好我仲係唔要	62
female	4.1	加上配音就有水從小管流出嚟	63
male	3.08	文革時唔理性嘅行為唔會重演	64
male	2.4	呢一位住客肯定唔簡單	65
male	1.64	亦都應該係咁樣	66
female	1.93	終於王舍新城	67
male	3.8	佢哋對政府嘅施政有啲乜嘢不滿意	68
male	1.51	幾昂貴嘅鋪租	69
male	2.22	人大常委唔應該插手	70
male	2.52	亦都難以明白玫瑰嘅心事	71
male	2.37	點樣掙扎都冇法子掙扎	72
male	9	大半都係借嚟過橋嘅啫不過咁到底都令人知道有啲自尊唔畀唔做事白討錢	73
male	1.24	向住某一個目的	74
male	3.29	有聲好書人性密碼修訂版	75
male	5.03	有所偏好誤入歧途故弄玄虛等等	76
female	2.2	生魚仲喺地上面跳	77
male	2.8	咁樣才能成為中國公民	78
female	1.36	我將門一開	79
female	1.86	呼朋三面坐下	80
female	3.19001	商務印書館香港有限公司出版	81
male	2.42	芙蓉如面柳如眉	82
female	2.36	實由於愛子情因所致	83
female	2.3	那麼鬼應再哭	84
female	6.51006	各自掏腰包有一次佢同早稻田大學教授實藤惠秀同行	85
male	2.84	手持象征權柄嘅大钺	86
female	1.91	大概有二三十個人	87
female	2.04	像經濟範疇	88
male	2.45	因為畢竟寫詩多年	89
female	1.97	佢嘅電單車就喺佢身後	90
male	0.99	就係呢一個	91
female	1.42	秀蘭功課好好啊	92
female	2.6	東去西去的典型旅途	93
female	2.19	我嘅一揮一斂	94
male	3.38	睇下呢一度嘅樹幾咁高幾咁直	95
female	4.42	嫩紅在我斑駁掌紋上顯得更玲瓏	96
female	2.63	仲有艇上人家傾倒每日嘅便溺	97
male	2.6	抵擋住急如雨下嘅警棍	98
male	5.5	七日裏邊有一日休息嘅勞工法例要到七十年代先至訂立	99

End of preview. Expand in Data Studio

MDCC: A New Cantonese ASR Dataset

📦 Update [1 Feb, 2024]

The .wav data of the dataset is available here:
🔗 Google Drive Link
Note: For research purposes only.

📖 Overview

MDCC (“Multi-Domain Cantonese Corpus”) is a large-scale Cantonese automatic speech recognition (ASR) dataset compiled from multiple domains. It provides:

Audio: .wav recordings of spontaneous and read speech
Transcript: UTF‑8 plain‑text transcripts
Speaker metadata: sex
Duration: audio length in seconds

This repo contains metadata files and a conversion script to turn the data into a Hugging Face-compatible dataset.

📝 Paper & Citation

Tiezheng Yu, Rita Frieske, Peng Xu, Samuel Cahyawijaya, Cheuk Tung Shadow Yiu, Holy Lovenia,
Wenliang Dai, Elham J. Barezi, Qifeng Chen, Xiaojuan Ma, Bertram E. Shi & Pascale Fung
“Automatic Speech Recognition Datasets in Cantonese: A Survey and New Dataset”
📄 arXiv:2201.02419

@misc{yu2022automatic,
  title        = {Automatic Speech Recognition Datasets in Cantonese: A Survey and New Dataset},
  author       = {Tiezheng Yu and Rita Frieske and Peng Xu and Samuel Cahyawijaya and
                  Cheuk Tung Shadow Yiu and Holy Lovenia and Wenliang Dai and
                  Elham J. Barezi and Qifeng Chen and Xiaojuan Ma and
                  Bertram E. Shi and Pascale Fung},
  year         = {2022},
  eprint       = {2201.02419},
  archivePrefix= {arXiv},
  primaryClass = {cs.CL}
}

🚀 How to Load on Hugging Face

from datasets import load_dataset

ds = load_dataset("ming030890/mdcc")
print(ds["train"][0])

Example output:

{
  'audio': {
    'path': '/path/to/audio.wav',
    'array': [...],
    'sampling_rate': 16000
  },
  'transcript': '你好，歡迎收聽…',
  'sex': 'female',
  'duration': 3.08
}

🔓 License & Access

Review the MDCC_LICENSE file in this repo.
Sign it and send to [email protected].
Then download the dataset here:
🔗 Google Drive Folder

✅ Checkpoints

Download pretrained models here:
🔗 Checkpoints Google Drive

⚠️ Disclaimer

I am not the original author of the dataset or the research paper.
This repo only provides a Hugging Face-compatible version of the public MDCC data.

For the original codebase and documentation, refer to:
🔗 https://github.com/HLTCHKUST/cantonese-asr

Downloads last month: 72

Size of downloaded dataset files:

14.7 GB

Size of the auto-converted Parquet files: