Datasets:
jimbozhang
commited on
Commit
•
dc01e20
1
Parent(s):
ed16922
Add word/phone level scores
Browse files- README.md +22 -5
- speechocean762.py +14 -0
README.md
CHANGED
@@ -42,10 +42,10 @@ pip install soundfile
|
|
42 |
|
43 |
>>> next(iter(test_set))
|
44 |
{'file': 'WAVE/SPEAKER0003/000030012.WAV',
|
45 |
-
'audio': {
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
'text': 'MARK IS GOING TO SEE ELEPHANT',
|
50 |
'speaker': '0003',
|
51 |
'gender': 'm',
|
@@ -53,7 +53,24 @@ pip install soundfile
|
|
53 |
'accuracy': 9,
|
54 |
'fluency': 9,
|
55 |
'prosodic': 9,
|
56 |
-
'total': 9
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
57 |
```
|
58 |
|
59 |
## The scoring metric
|
|
|
42 |
|
43 |
>>> next(iter(test_set))
|
44 |
{'file': 'WAVE/SPEAKER0003/000030012.WAV',
|
45 |
+
'audio': {
|
46 |
+
'path': 'WAVE/SPEAKER0003/000030012.WAV',
|
47 |
+
'array': array([-0.00119019, -0.00500488, -0.00283813, ..., 0.00274658, 0. , 0.00125122]),
|
48 |
+
'sampling_rate': 16000},
|
49 |
'text': 'MARK IS GOING TO SEE ELEPHANT',
|
50 |
'speaker': '0003',
|
51 |
'gender': 'm',
|
|
|
53 |
'accuracy': 9,
|
54 |
'fluency': 9,
|
55 |
'prosodic': 9,
|
56 |
+
'total': 9,
|
57 |
+
'words': {'text': ['MARK', 'IS', 'GOING', 'TO', 'SEE', 'ELEPHANT'],
|
58 |
+
'accuracy': [10, 10, 10, 10, 10, 10],
|
59 |
+
'stress': [10, 10, 10, 10, 10, 10],
|
60 |
+
'total': [10, 10, 10, 10, 10, 10],
|
61 |
+
'phones': [['M', 'AA0', 'R', 'K'],
|
62 |
+
['IH0', 'Z'],
|
63 |
+
['G', 'OW0', 'IH0', 'NG'],
|
64 |
+
['T', 'UW0'],
|
65 |
+
['S', 'IY0'],
|
66 |
+
['EH1', 'L', 'IH0', 'F', 'AH0', 'N', 'T']],
|
67 |
+
'phones-accuracy': [[2.0, 2.0, 1.8, 2.0],
|
68 |
+
[2.0, 1.8],
|
69 |
+
[2.0, 2.0, 2.0, 2.0],
|
70 |
+
[2.0, 2.0],
|
71 |
+
[2.0, 2.0],
|
72 |
+
[2.0, 2.0, 2.0, 2.0, 2.0, 2.0, 2.0]],
|
73 |
+
'mispronunciations': ['[]', '[]', '[]', '[]', '[]', '[]']}}
|
74 |
```
|
75 |
|
76 |
## The scoring metric
|
speechocean762.py
CHANGED
@@ -71,6 +71,19 @@ class Speechocean762(datasets.GeneratorBasedBuilder):
|
|
71 |
"fluency": datasets.Value("int16"),
|
72 |
"prosodic": datasets.Value("int16"),
|
73 |
"total": datasets.Value("int16"),
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
}
|
75 |
),
|
76 |
homepage=_HOMEPAGE,
|
@@ -110,4 +123,5 @@ class Speechocean762(datasets.GeneratorBasedBuilder):
|
|
110 |
"fluency": row["fluency"],
|
111 |
"prosodic": row["prosodic"],
|
112 |
"total": row["total"],
|
|
|
113 |
}
|
|
|
71 |
"fluency": datasets.Value("int16"),
|
72 |
"prosodic": datasets.Value("int16"),
|
73 |
"total": datasets.Value("int16"),
|
74 |
+
"words": datasets.Sequence(
|
75 |
+
feature={
|
76 |
+
"text": datasets.Value("string"),
|
77 |
+
"accuracy": datasets.Value("int16"),
|
78 |
+
"stress": datasets.Value("int16"),
|
79 |
+
"total": datasets.Value("int16"),
|
80 |
+
"phones": datasets.Sequence(datasets.Value("string")),
|
81 |
+
"phones-accuracy": datasets.Sequence(
|
82 |
+
datasets.Value("float")
|
83 |
+
),
|
84 |
+
"mispronunciations": datasets.Value("string"),
|
85 |
+
}
|
86 |
+
),
|
87 |
}
|
88 |
),
|
89 |
homepage=_HOMEPAGE,
|
|
|
123 |
"fluency": row["fluency"],
|
124 |
"prosodic": row["prosodic"],
|
125 |
"total": row["total"],
|
126 |
+
"words": row["words"],
|
127 |
}
|