ihanif commited on
Commit
a20e641
Β·
verified Β·
1 Parent(s): 19edb28

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +107 -0
README.md ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - ps
4
+ license: mit
5
+ pretty_name: Pashto Synthetic Speech Dataset Parquet (3k)
6
+ size_categories:
7
+ - 1K<n<10K
8
+ tags:
9
+ - audio
10
+ - speech
11
+ - pashto
12
+ - tts
13
+ - synthetic
14
+ - parquet
15
+ task_categories:
16
+ - automatic-speech-recognition
17
+ metrics:
18
+ - wer
19
+ paperswithcode:
20
+ eval_results: []
21
+ preferred_citation: |-
22
+ @dataset{pashto_synthetic_speech_parquet_3k,
23
+ title = {Pashto Synthetic Speech Dataset Parquet (3k)},
24
+ year = {2023}
25
+ }
26
+ ---
27
+
28
+ # Pashto Synthetic Speech Dataset Parquet (3k)
29
+
30
+ This dataset contains 6000 synthetic speech recordings in the Pashto language,
31
+ with 3000 male voice recordings and 3000 female voice recordings, stored in Parquet format.
32
+
33
+ ## Dataset Information
34
+
35
+ - **Dataset Size**: 3000 sentences
36
+ - **Total Recordings**: 6000 audio files (3000 male + 3000 female)
37
+ - **Audio Format**: WAV, 44.1kHz, 16-bit PCM, embedded directly in Parquet files
38
+ - **Dataset Format**: Parquet with 500MB shards
39
+ - **Sampling Rate**: 44.1kHz (44100 Hz)
40
+ - **Voices**:
41
+ - Male: ps-AF-GulNawazNeural
42
+ - Female: ps-AF-LatifaNeural
43
+ - **Total Audio Duration**:
44
+ - Male: 1596.27 seconds
45
+ - Female: 1635.43 seconds
46
+ - **Average Clip Duration**:
47
+ - Male: 0.53 seconds
48
+ - Female: 0.55 seconds
49
+
50
+ ## Dataset Structure
51
+
52
+ ```
53
+ pashto_speech_parquet_3k/
54
+ β”œβ”€β”€ data/ # Parquet data files with embedded audio
55
+ β”‚ β”œβ”€β”€ shard-00000-of-....parquet
56
+ β”‚ β”œβ”€β”€ shard-00001-of-....parquet
57
+ β”‚ β”œβ”€β”€ ...
58
+ β”‚ β”œβ”€β”€ _schema.json # Parquet schema definition
59
+ β”‚ └── _index.json # Parquet metadata
60
+ β”œβ”€β”€ dataset_info.json # Dataset metadata
61
+ └── README.md # This file
62
+ ```
63
+
64
+ ## Usage with HuggingFace Datasets
65
+
66
+ ```python
67
+ from datasets import load_dataset
68
+
69
+ # Load the dataset
70
+ dataset = load_dataset("your-username/pashto_speech_parquet_3k")
71
+
72
+ # Access an example
73
+ example = dataset['train'][0]
74
+ print(example['text']) # The transcript
75
+ print(example['audio']) # Audio data
76
+ print(example['speaker']) # Speaker gender
77
+ print(example['sampling_rate']) # Audio sampling rate
78
+ ```
79
+
80
+ ## Usage Applications
81
+
82
+ This dataset can be used for:
83
+ - Training ASR (Automatic Speech Recognition) systems for Pashto
84
+ - Voice activity detection
85
+ - Speaker verification
86
+ - Speech synthesis evaluation
87
+
88
+ ## License
89
+
90
+ This dataset is licensed under the MIT License.
91
+
92
+ ## Citation
93
+
94
+ If you use this dataset in your research, please cite:
95
+
96
+ ```
97
+ @dataset{pashto_synthetic_speech_parquet_3k,
98
+ title = {Pashto Synthetic Speech Dataset Parquet (3k)},
99
+ year = {2023},
100
+ publisher = {HuggingFace},
101
+ howpublished = {\url{https://huggingface.co/datasets/ihanif/pashto_speech_parquet_3k}}
102
+ }
103
+ ```
104
+
105
+ ## Acknowledgments
106
+
107
+ This dataset was created using Microsoft Edge TTS voices.