lazy-worm commited on
Commit
02e1654
·
verified ·
1 Parent(s): 0175160

Upload folder using huggingface_hub

Browse files
README.md ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - nan
4
+ license: other
5
+ task_categories:
6
+ - automatic-speech-recognition
7
+ tags:
8
+ - speech
9
+ - audio
10
+ - asr
11
+ ---
12
+
13
+ # lai-ching-te-speech
14
+
15
+ ASR dataset packaged from SRT-aligned audio clips.
16
+
17
+ ## Dataset info
18
+
19
+ | Split | Samples |
20
+ |------------|---------|
21
+ | train | 10 |
22
+ | validation | 0 |
23
+ | test | 0 |
24
+
25
+ Total audio: **0.0 hours**
26
+
27
+ ## Usage
28
+
29
+ ```python
30
+ from datasets import load_dataset, Audio
31
+
32
+ ds = load_dataset("lai-ching-te-speech")
33
+ ds = ds.cast_column("audio", Audio(sampling_rate=16000))
34
+
35
+ # Access a sample
36
+ sample = ds["train"][0]
37
+ print(sample["text"]) # transcript
38
+ print(sample["audio"]) # {"array": ..., "sampling_rate": ...}
39
+ ```
40
+
41
+ ## Columns
42
+
43
+ - `audio` — raw WAV bytes embedded in Parquet
44
+ - `text` — ASR-normalized transcript (lowercase, no punctuation)
45
+ - `duration_s` — clip duration in seconds
46
+ - `snr_db` — estimated signal-to-noise ratio
47
+ - `wps` — words per second (speaking rate)
48
+ - `original_text` — raw subtitle text before normalization
dataset_info.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset_name": "lai-ching-te-speech",
3
+ "mode": "ASR",
4
+ "splits": {
5
+ "train": 10
6
+ },
7
+ "total_samples": 10,
8
+ "total_hours": 0.03,
9
+ "sample_rate": 16000,
10
+ "schema": [
11
+ "audio",
12
+ "text",
13
+ "duration_s",
14
+ "snr_db",
15
+ "wps",
16
+ "original_text",
17
+ "split"
18
+ ]
19
+ }
metadata_5_packaged.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "stage": 5,
3
+ "name": "packager",
4
+ "args": {
5
+ "splits": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized\\splits",
6
+ "wav_dir": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized",
7
+ "out": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\packaged",
8
+ "tts": false,
9
+ "speaker_col": null,
10
+ "shard_size": 500,
11
+ "dataset_name": "lai-ching-te-speech",
12
+ "push_to_hub": null,
13
+ "token": null,
14
+ "lang": "nan"
15
+ },
16
+ "stats": {
17
+ "dataset_name": "lai-ching-te-speech",
18
+ "mode": "ASR",
19
+ "splits": {
20
+ "train": 10
21
+ },
22
+ "total_samples": 10,
23
+ "total_hours": 0.03,
24
+ "sample_rate": 16000,
25
+ "schema": [
26
+ "audio",
27
+ "text",
28
+ "duration_s",
29
+ "snr_db",
30
+ "wps",
31
+ "original_text",
32
+ "split"
33
+ ]
34
+ },
35
+ "subtitle_metadata": {
36
+ "Sync Offset": "203.500s",
37
+ "Dropped Gaps": "[]",
38
+ "Title": "\u5168\u6587\uff0f\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\uff01\u8d8a\u8b1b\u8d8a\u6fc0\u6602\u3000\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b\u6c7a\u5fc3\uff0d\u6c11\u8996\u65b0\u805e",
39
+ "Channel": "\u6c11\u8996\u65b0\u805e\u7db2 Formosa TV News network",
40
+ "Video ID": "OtZUjdHpuWM",
41
+ "URL": "https://www.youtube.com/watch?v=OtZUjdHpuWM",
42
+ "Language": "Taigi",
43
+ "Duration": "883s",
44
+ "Total Segments": "21"
45
+ }
46
+ }
metadata_overall.json ADDED
@@ -0,0 +1,227 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "pipeline_args": {
3
+ "start_stage": 1,
4
+ "end_stage": 5,
5
+ "skip_normalization": false,
6
+ "clips": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\media-downloads\\video\\clips\\local_cut_\u5168\u6587\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\u8d8a\u8b1b\u8d8a\u6fc0\u6602_\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b_204.0_300.0_20260316_140632",
7
+ "manifest": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\media-downloads\\video\\clips\\local_cut_\u5168\u6587\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\u8d8a\u8b1b\u8d8a\u6fc0\u6602_\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b_204.0_300.0_20260316_140632\\local_cut_\u5168\u6587\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\u8d8a\u8b1b\u8d8a\u6fc0\u6602_\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b_204.0_300.0_20260316_140632_manifest.csv",
8
+ "dataset_name": "lai-ching-te-speech",
9
+ "norm_dir": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized",
10
+ "out": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\packaged",
11
+ "text_col": null,
12
+ "sr": 16000,
13
+ "lufs": -23.0,
14
+ "workers": 4,
15
+ "tts": false,
16
+ "lang": "nan",
17
+ "max_tts_chars": 200,
18
+ "remove_punctuation": false,
19
+ "cer": false,
20
+ "filter_audio": true,
21
+ "min_dur": null,
22
+ "max_dur": null,
23
+ "min_snr": 20.0,
24
+ "max_cer": 0.15,
25
+ "min_wps": 0.0,
26
+ "max_wps": 5.0,
27
+ "train": 0.9,
28
+ "val": 0.05,
29
+ "test": 0.05,
30
+ "seed": 42,
31
+ "speaker_col": null,
32
+ "stratify_dur": false,
33
+ "shard_size": 500,
34
+ "push_to_hub": null,
35
+ "token": null
36
+ },
37
+ "subtitle_metadata": {
38
+ "Sync Offset": "203.500s",
39
+ "Dropped Gaps": "[]",
40
+ "Title": "\u5168\u6587\uff0f\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\uff01\u8d8a\u8b1b\u8d8a\u6fc0\u6602\u3000\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b\u6c7a\u5fc3\uff0d\u6c11\u8996\u65b0\u805e",
41
+ "Channel": "\u6c11\u8996\u65b0\u805e\u7db2 Formosa TV News network",
42
+ "Video ID": "OtZUjdHpuWM",
43
+ "URL": "https://www.youtube.com/watch?v=OtZUjdHpuWM",
44
+ "Language": "Taigi",
45
+ "Duration": "883s",
46
+ "Total Segments": "21"
47
+ },
48
+ "stages": {
49
+ "1_normalized": {
50
+ "stage": 1,
51
+ "name": "audio_normalizer",
52
+ "args": {
53
+ "clips": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\media-downloads\\video\\clips\\local_cut_\u5168\u6587\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\u8d8a\u8b1b\u8d8a\u6fc0\u6602_\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b_204.0_300.0_20260316_140632",
54
+ "out": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized",
55
+ "sr": 16000,
56
+ "lufs": -23.0,
57
+ "workers": 4,
58
+ "manifest": "local_cut_\u5168\u6587\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\u8d8a\u8b1b\u8d8a\u6fc0\u6602_\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b_204.0_300.0_20260316_140632_manifest.csv"
59
+ },
60
+ "stats": {
61
+ "normalized": 11,
62
+ "errors": 0
63
+ },
64
+ "subtitle_metadata": {
65
+ "Sync Offset": "203.500s",
66
+ "Dropped Gaps": "[]",
67
+ "Title": "\u5168\u6587\uff0f\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\uff01\u8d8a\u8b1b\u8d8a\u6fc0\u6602\u3000\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b\u6c7a\u5fc3\uff0d\u6c11\u8996\u65b0\u805e",
68
+ "Channel": "\u6c11\u8996\u65b0\u805e\u7db2 Formosa TV News network",
69
+ "Video ID": "OtZUjdHpuWM",
70
+ "URL": "https://www.youtube.com/watch?v=OtZUjdHpuWM",
71
+ "Language": "Taigi",
72
+ "Duration": "883s",
73
+ "Total Segments": "21"
74
+ }
75
+ },
76
+ "2_cleaned": {
77
+ "stage": 2,
78
+ "name": "text_cleaner",
79
+ "args": {
80
+ "manifest": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized\\_manifest_1_normalized.csv",
81
+ "tts": false,
82
+ "lang": "nan",
83
+ "max_tts_chars": 200,
84
+ "remove_punctuation": false
85
+ },
86
+ "stats": {
87
+ "kept": 11,
88
+ "dropped": 0
89
+ },
90
+ "subtitle_metadata": {
91
+ "Sync Offset": "203.500s",
92
+ "Dropped Gaps": "[]",
93
+ "Title": "\u5168\u6587\uff0f\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\uff01\u8d8a\u8b1b\u8d8a\u6fc0\u6602\u3000\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b\u6c7a\u5fc3\uff0d\u6c11\u8996\u65b0\u805e",
94
+ "Channel": "\u6c11\u8996\u65b0\u805e\u7db2 Formosa TV News network",
95
+ "Video ID": "OtZUjdHpuWM",
96
+ "URL": "https://www.youtube.com/watch?v=OtZUjdHpuWM",
97
+ "Language": "Taigi",
98
+ "Duration": "883s",
99
+ "Total Segments": "21"
100
+ }
101
+ },
102
+ "3_filtered": {
103
+ "stage": 3,
104
+ "name": "quality_filter",
105
+ "args": {
106
+ "manifest": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized\\_manifest_2_cleaned.csv",
107
+ "text_col": null,
108
+ "tts": false,
109
+ "cer": false,
110
+ "min_dur": null,
111
+ "max_dur": null,
112
+ "min_snr": 20.0,
113
+ "max_cer": 0.15,
114
+ "min_wps": 0.0,
115
+ "max_wps": 5.0,
116
+ "filter_audio": true,
117
+ "wav_dir": null,
118
+ "device": null
119
+ },
120
+ "stats": {
121
+ "kept": 10,
122
+ "dropped": 1,
123
+ "reasons": {
124
+ "snr": 1
125
+ },
126
+ "audio_checks_skipped": false
127
+ },
128
+ "subtitle_metadata": {
129
+ "Sync Offset": "203.500s",
130
+ "Dropped Gaps": "[]",
131
+ "Title": "\u5168\u6587\uff0f\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\uff01\u8d8a\u8b1b\u8d8a\u6fc0\u6602\u3000\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b\u6c7a\u5fc3\uff0d\u6c11\u8996\u65b0\u805e",
132
+ "Channel": "\u6c11\u8996\u65b0\u805e\u7db2 Formosa TV News network",
133
+ "Video ID": "OtZUjdHpuWM",
134
+ "URL": "https://www.youtube.com/watch?v=OtZUjdHpuWM",
135
+ "Language": "Taigi",
136
+ "Duration": "883s",
137
+ "Total Segments": "21"
138
+ }
139
+ },
140
+ "4_split": {
141
+ "stage": 4,
142
+ "name": "split_dataset",
143
+ "args": {
144
+ "manifest": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized\\_manifest_3_filtered.csv",
145
+ "train": 0.9,
146
+ "val": 0.05,
147
+ "test": 0.05,
148
+ "seed": 42,
149
+ "speaker_col": null,
150
+ "stratify_dur": false,
151
+ "out": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized\\splits"
152
+ },
153
+ "stats": {
154
+ "total": 10,
155
+ "train": 10,
156
+ "val": 0,
157
+ "test": 0,
158
+ "train_ratio": 1.0,
159
+ "val_ratio": 0.0,
160
+ "test_ratio": 0.0,
161
+ "seed": 42,
162
+ "method": "random"
163
+ },
164
+ "subtitle_metadata": {
165
+ "Sync Offset": "203.500s",
166
+ "Dropped Gaps": "[]",
167
+ "Title": "\u5168\u6587\uff0f\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\uff01\u8d8a\u8b1b\u8d8a\u6fc0\u6602\u3000\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b\u6c7a\u5fc3\uff0d\u6c11\u8996\u65b0\u805e",
168
+ "Channel": "\u6c11\u8996\u65b0\u805e\u7db2 Formosa TV News network",
169
+ "Video ID": "OtZUjdHpuWM",
170
+ "URL": "https://www.youtube.com/watch?v=OtZUjdHpuWM",
171
+ "Language": "Taigi",
172
+ "Duration": "883s",
173
+ "Total Segments": "21"
174
+ }
175
+ },
176
+ "5_packaged": {
177
+ "stage": 5,
178
+ "name": "packager",
179
+ "args": {
180
+ "splits": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized\\splits",
181
+ "wav_dir": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\normalized",
182
+ "out": "C:\\Users\\marti\\Projects\\taigi_voice_datasets_packager\\datasets\\lai-ching-te-speech\\packaged",
183
+ "tts": false,
184
+ "speaker_col": null,
185
+ "shard_size": 500,
186
+ "dataset_name": "lai-ching-te-speech",
187
+ "push_to_hub": null,
188
+ "token": null,
189
+ "lang": "nan"
190
+ },
191
+ "stats": {
192
+ "dataset_name": "lai-ching-te-speech",
193
+ "mode": "ASR",
194
+ "splits": {
195
+ "train": 10
196
+ },
197
+ "total_samples": 10,
198
+ "total_hours": 0.03,
199
+ "sample_rate": 16000,
200
+ "schema": [
201
+ "audio",
202
+ "text",
203
+ "duration_s",
204
+ "snr_db",
205
+ "wps",
206
+ "original_text",
207
+ "split"
208
+ ]
209
+ },
210
+ "subtitle_metadata": {
211
+ "Sync Offset": "203.500s",
212
+ "Dropped Gaps": "[]",
213
+ "Title": "\u5168\u6587\uff0f\u8cf4\u6e05\u5fb7\u5168\u4ee3\u6703\u6f14\u8aaa\u5167\u5bb9\u4e00\u6b21\u770b\uff01\u8d8a\u8b1b\u8d8a\u6fc0\u6602\u3000\u5168\u7a0b\u53f0\u8a9e\u8868\u9054\u8b77\u6c11\u4e3b\u6c7a\u5fc3\uff0d\u6c11\u8996\u65b0\u805e",
214
+ "Channel": "\u6c11\u8996\u65b0\u805e\u7db2 Formosa TV News network",
215
+ "Video ID": "OtZUjdHpuWM",
216
+ "URL": "https://www.youtube.com/watch?v=OtZUjdHpuWM",
217
+ "Language": "Taigi",
218
+ "Duration": "883s",
219
+ "Total Segments": "21"
220
+ }
221
+ }
222
+ },
223
+ "start_time": 1774745448.6587112,
224
+ "success": true,
225
+ "end_time": 1774745452.893064,
226
+ "elapsed_s": 4.2343528270721436
227
+ }
stats.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Dataset Statistics
2
+
3
+ | Dataset | Path | Description | Count | Total Hours | Min (s) | Max (s) | Mean (s) | P50 (s) | P75 (s) | P90 (s) | P95 (s) | P99 (s) |
4
+ |---|---|---|---|---|---|---|---|---|---|---|---|---|
5
+ | lai-ching-te-speech | packaged | lai-ching-te-speech Dataset | 10 | 0.03 | 6.13 | 14.12 | 9.03 | 8.13 | 10.87 | 14.12 | 14.12 | 14.12 |
6
+
7
+ # Random Text Examples
8
+
9
+ ## lai-ching-te-speech (10 examples)
10
+
11
+ 1. (現場 pho̍k-á-siaⁿ)
12
+ 2. 一方面突破戒嚴,拍破黨禁、報禁,追求百分之百 ê 言論自由。
13
+ 3. 由執政淪為在野,又 koh 再一次執政。
14
+ 4. 今年正月份 ê 總統大選,台灣人民拍開歷史 ê 新頁,連紲第三遍將國家 ê 重擔交 hō͘ 民進黨。
15
+ 5. 另外一方面推動國會全面改選、總統直選,帶動國家民主進步、社會多元發展。
16
+ 6. 秉持清廉、勤政、愛鄉土 ê 精神,和人民徛做伙。
17
+ 7. 咱民進黨,誕生 tī 專制獨裁年代,
18
+ 8. 這个過程中,民進黨 mā 由在野到執政,
19
+ 9. 就是期待咱會當繼續帶領台灣,行對 ê 路,深化民主憲政,全力鞏固和平 kap 繁榮。
20
+ 10. 所以今仔日,咱有機會 tī 遮召開全國黨員代表大會,咱應該愛用感謝先輩,in 用血汗、用性命,àn 威權拚出台灣 ê 民主 kap 自由,
21
+
train/data-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ebb921c907e383e0f5732acb4b1e18e7aed0b7ed175205e7711c076915bbe7bf
3
+ size 2885024