Datasets:
Tags:
speech-modeling
License:
Update README.md
Browse files
README.md
CHANGED
@@ -61,22 +61,6 @@ data = load_dataset("NbAiLab/NPSC", config="16K_mp3_bokmaal", streaming=True)
|
|
61 |
### Dataset Summary
|
62 |
The NPSC dataset contains json lines with language training data. Here is an example json line:
|
63 |
```json
|
64 |
-
{
|
65 |
-
"meeting_date": datasets.Value("string"),
|
66 |
-
"sentence_order": datasets.Value("int32"),
|
67 |
-
"speaker_id" : datasets.Value("int32"),
|
68 |
-
"speaker_name": datasets.Value("string"),
|
69 |
-
"sentence_text": datasets.Value("string"),
|
70 |
-
"sentence_language_code": datasets.Value("string"),
|
71 |
-
"text": datasets.Value("string"),
|
72 |
-
"start_time": datasets.Value("int32"),
|
73 |
-
"end_time": datasets.Value("int32"),
|
74 |
-
"normsentence_text": datasets.Value("string"),
|
75 |
-
"transsentence_text": datasets.Value("string"),
|
76 |
-
"translated": datasets.Value("int32"),
|
77 |
-
"audio": datasets.features.Audio(sampling_rate=sampling_rate),
|
78 |
-
|
79 |
-
}
|
80 |
{
|
81 |
"sentence_id": 49853,
|
82 |
"sentence_order": 0,
|
@@ -85,14 +69,12 @@ The NPSC dataset contains json lines with language training data. Here is an exa
|
|
85 |
"sentence_text": "Stortingets møte er lovlig satt",
|
86 |
"sentence_language_code": "nb-NO",
|
87 |
"text": "Stortingets møte er lovlig satt",
|
88 |
-
"start_time": 320246,
|
|
|
89 |
"normsentence_text": "Stortingets møte er lovlig satt",
|
90 |
"transsentence_text": "Stortingets møte er lovleg sett",
|
91 |
"translated": 1,
|
92 |
-
"audio": {"path": "audio/20170110-095504_320246_323590.wav",
|
93 |
-
"array": [.......]
|
94 |
-
}
|
95 |
-
|
96 |
}
|
97 |
!!!PES: Not all the fields above and below correspond to what is actually found
|
98 |
in the json files. It would be good if you could correct this. When you have, I can
|
@@ -100,25 +82,26 @@ add further descriptions of the fields where needed.!!!
|
|
100 |
|
101 |
```
|
102 |
## Data Fields (!!!!We really need a descrption here - and maybe some cleaning!!!)
|
103 |
-
|**
|
104 |
-
|
105 |
-
|**
|
106 |
-
|**
|
107 |
-
|
108 |
-
| **
|
109 |
-
| **
|
110 |
-
| **
|
111 |
-
| **
|
112 |
-
| **
|
113 |
-
| **
|
114 |
-
| **
|
115 |
-
| **
|
116 |
-
| **
|
|
|
117 |
|
118 |
|
119 |
|
120 |
### Dataset Creation
|
121 |
-
We are providing a **train** and a **validation** split. The standard size of the validation is a single 1GB file, while train is
|
122 |
All files are gzipped. There is also a **test** split, which is not shared here. It is, however part of the [official release of the NPSC](https://www.nb.no/sprakbanken/ressurskatalog/oai-nb-no-sbr-58/).
|
123 |
|
124 |
|
@@ -139,9 +122,6 @@ The procedure for the dataset creation is described in detail in our paper.
|
|
139 |
| | Male: 61.7% |
|
140 |
|
141 |
|
142 |
-
!!!It would be great to know how many hours of Bokmål/Nynorsk/English(?) here!!!!!
|
143 |
-
!!!PES: As long as we have the percentages, I don't think it is necessary!!!
|
144 |
-
|
145 |
## Considerations for Using the Data
|
146 |
This corpus contains speech data. All recordings are of members of Parliament in a public setting, and can be distributed without any restrains.
|
147 |
|
|
|
61 |
### Dataset Summary
|
62 |
The NPSC dataset contains json lines with language training data. Here is an example json line:
|
63 |
```json
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
64 |
{
|
65 |
"sentence_id": 49853,
|
66 |
"sentence_order": 0,
|
|
|
69 |
"sentence_text": "Stortingets møte er lovlig satt",
|
70 |
"sentence_language_code": "nb-NO",
|
71 |
"text": "Stortingets møte er lovlig satt",
|
72 |
+
"start_time": 320246,
|
73 |
+
"end_time": 323590,
|
74 |
"normsentence_text": "Stortingets møte er lovlig satt",
|
75 |
"transsentence_text": "Stortingets møte er lovleg sett",
|
76 |
"translated": 1,
|
77 |
+
"audio": {"path": "audio/20170110-095504_320246_323590.wav","array": [.......]}
|
|
|
|
|
|
|
78 |
}
|
79 |
!!!PES: Not all the fields above and below correspond to what is actually found
|
80 |
in the json files. It would be good if you could correct this. When you have, I can
|
|
|
82 |
|
83 |
```
|
84 |
## Data Fields (!!!!We really need a descrption here - and maybe some cleaning!!!)
|
85 |
+
|**Key** | **Type** | **Description** |
|
86 |
+
|:-----------|:------------|:------------|
|
87 |
+
|**sentence_id:** | Integer | String with id to source of line and a unique identifier|
|
88 |
+
|**sentence_order** | String | with order of sentence |
|
89 |
+
|**speaker id** | Integer | id of speaker |
|
90 |
+
| **speaker_name** | String | name of speaker |
|
91 |
+
| **sentence_text** | String | sentence text |
|
92 |
+
| **sentence_language_code** | String | sentence text (!!!LIST ALL ALTERNATIVES!!!)|
|
93 |
+
| **text** | String | sentence text. This is a copy of "sentence_text". It is included here to make it more convenient to interleave with other datasets.|
|
94 |
+
| **start_time** | Integer | start time |
|
95 |
+
| **end_time** | Integer | end time |
|
96 |
+
| **normsentence_text** | String | normalised sentence text |
|
97 |
+
| **transsentence_text** | String | translated sentence text |
|
98 |
+
| **translated** | Integer | text translated |
|
99 |
+
| **audio** | array | Encoded audio with 'path', 'array','sampling_rate' |
|
100 |
|
101 |
|
102 |
|
103 |
### Dataset Creation
|
104 |
+
We are providing a **train** and a **validation** split. The standard size of the validation is a single 1GB file, while train is charded in 1GB chunks.
|
105 |
All files are gzipped. There is also a **test** split, which is not shared here. It is, however part of the [official release of the NPSC](https://www.nb.no/sprakbanken/ressurskatalog/oai-nb-no-sbr-58/).
|
106 |
|
107 |
|
|
|
122 |
| | Male: 61.7% |
|
123 |
|
124 |
|
|
|
|
|
|
|
125 |
## Considerations for Using the Data
|
126 |
This corpus contains speech data. All recordings are of members of Parliament in a public setting, and can be distributed without any restrains.
|
127 |
|