jayliqinzhang commited on
Commit
92c9829
·
verified ·
1 Parent(s): 4f3b886

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +136 -3
README.md CHANGED
@@ -1,3 +1,136 @@
1
- ---
2
- license: cc-by-4.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc0-1.0
3
+ task_categories:
4
+ - text-to-speech
5
+ - automatic-speech-recognition
6
+ language:
7
+ - en
8
+ - de
9
+ - fr
10
+ - bg
11
+ - ar
12
+
13
+ pretty_name: Mumospee_small
14
+ tags:
15
+ - Speech
16
+ - Video
17
+ ---
18
+
19
+ # Mumospee: A MUltiMOdal SPEEch Corpus (small)
20
+
21
+ This is a small version of Mumospee include no more 1000 rows. The full is [here](https://huggingface.co/datasets/meetween/mumospee).
22
+
23
+ ## Overview
24
+
25
+ At this version it includes below languages and data soruces:
26
+
27
+ ```python
28
+ _LANGUAGES = ["en", "bg", "de", "ar", "fr"]
29
+ _TAGS = ["CoVoST", "GigaSpeech", "PeopleSpeech", "Librispeech", "LibriTTS", "Emilia", "MOSEL"]
30
+ ```
31
+
32
+
33
+ ## Data Sources
34
+ The initial release includes metadata and download scripts for accessing the following publicly available datasets:
35
+
36
+ - [CoVoST](https://github.com/facebookresearch/covost)
37
+ - [GigaSpeech](https://github.com/SpeechColab/GigaSpeech)
38
+ - [people-speech](https://mlcommons.org/datasets/peoples-speech/)
39
+ - [LibriSpeech](https://www.openslr.org/12)
40
+ - [LibriTTS](https://openslr.org/60/)
41
+ - [Emilia](https://emilia-dataset.github.io/Emilia-Demo-Page/#dataset)
42
+ - [MOSEL](https://huggingface.co/datasets/FBK-MT/mosel)
43
+
44
+
45
+ ## Mumospee dataset structure
46
+
47
+ Mumospee is available at [HuggingFace](https://huggingface.co/datasets/meetween/mumospee) without providing all the audio data directly, but the urls or scripts to access the datasets.
48
+
49
+ In the metadata csv, each row is a sample representing the metadata of an audio, a video or a clip consisting of the following information:
50
+
51
+
52
+ - "path": the relative path of the audio file to the sample.
53
+ - "url": the link to download the parquet containing the audio, video or the clip of it.
54
+ - "type": the sample is an audio or video.
55
+ - "duration": the duration of the sample in second.
56
+ - "language": the language of the video or audio.
57
+ - "transcript": the transcript of the video or audio.
58
+ - "tag": the origin of the sample.
59
+ - "split": the sample is in split, test, or validation section in the original dataset.
60
+ - "license": the license to use this sample.
61
+
62
+
63
+ ```json
64
+ {
65
+ "path": "3660-172183-0000.flac",
66
+ "url": "https://huggingface.co/datasets/meetween/mumospee_librispeech/resolve/main/librispeech-parquet/dev-other.parquet",
67
+ "type": "audio",
68
+ "duration": 5.405,
69
+ "language": "en",
70
+ "transcript": "GERAINT AS HE HAD BEEN USED TO DO WHEN HE WAS AT ARTHUR'S COURT FREQUENTED TOURNAMENTS",
71
+ "tag": "Librispeech",
72
+ "split": "validation",
73
+ "license": "CC-BY-4.0"
74
+ }
75
+ ```
76
+
77
+
78
+ ## Example Usage
79
+
80
+
81
+ ```python
82
+
83
+ dataset= load_dataset("meetween/mumospee_small", trust_remote_code=True)
84
+ print(dataset)
85
+
86
+ # To get the first row of the dataset.
87
+ sample_first = dataset["train"][0]
88
+ ```
89
+ The defaul outputs all the samples from train split. To get another splits:
90
+
91
+ ```python
92
+
93
+ # To get the dataset from test or validation split
94
+ dataset_test = load_dataset("meetween/mumospee_small", "test", trust_remote_code=True)
95
+ dataset_validation = load_dataset("meetween/mumospee_small", "validation", trust_remote_code=True)
96
+ ```
97
+
98
+ - ### Filters ###
99
+
100
+ There are filters to select dataset samples from specific groups:
101
+
102
+ ```python
103
+
104
+ # To get the dataset of langauge "en".
105
+ dataset= load_dataset("meetween/mumospee_small", "test", language="en", trust_remote_code=True)
106
+
107
+ # To get the dataset from MOSEL.
108
+ dataset= load_dataset("meetween/mumospee_small", "train", tag="MOSEL", trust_remote_code=True)
109
+
110
+ # You can also add combination of language and tag: get English from CoVoST from test split.
111
+ dataset= load_dataset("meetween/mumospee_small", "test", language="en", tag="CoVoST", trust_remote_code=True)
112
+
113
+ ```
114
+
115
+ Note: keep in mind that if a filter combination (including split) results to no dataset, you may get an value error like below:
116
+
117
+ ```python
118
+ ValueError: Instruction "train" corresponds to no data!
119
+ ```
120
+
121
+ Also, make sure the values are from `_LANGUAGES` and `_TAG`.
122
+
123
+ - ### Download audios ###
124
+
125
+ You can download the parquet files with the audios data by using the `download_audio` parameter (the default is `None`):
126
+
127
+ ```python
128
+ dataset= load_dataset("meetween/mumospee_small", "test", download_audio=True, language="en", trust_remote_code=True)
129
+ ```
130
+
131
+
132
+ ## License
133
+
134
+ The metadata and download scripts are publicly available under a CC0 license. While the metadata itself is open, users must comply with the licensing terms of each underlying dataset.
135
+
136
+ ---