Upload README (1).md

README (1).md

@@ -0,0 +1,64 @@
+# 1. Data Cleaning and Validation:
+- Samples containing Ruby tag as well as ■ are removed by using regex on {} and ruby ( the amount of samples removed nearly 100 samples)
+- audio_path/genshin/genshin_english/audio/"The_Captain"_112313.wav: this type of audio can run, I check by using this code:
+
+```python
+import os
+import librosa
+import matplotlib.pyplot as plt
+import numpy as np
+if os.path.exists(path):
+
+    # Load the audio and resample to 24,000 Hz
+    wave, sr = librosa.load(path, sr=24000)
+
+# Print basic info
+print(f"Sample rate: {sr}")
+print(f"Waveform shape: {wave.shape}")
+
+print(f"Duration: {len(wave) / sr:.2f} seconds")
+
+# Optional: Plot the waveform
+plt.figure(figsize=(10, 4))
+plt.plot(np.linspace(0, len(wave) / sr, num=len(wave)), wave)
+plt.title("Waveform")
+plt.xlabel("Time (s)")
+plt.ylabel("Amplitude")
+plt.tight_layout()
+plt.show()
+```
+
+# 2. CSV and Phoneme Files:
+- all_EN_game_v2.csv: this is the final version containing path,transcript, speaker_id (not number). Game are wutheringWaves_Dataset_new, honkai_starail and genshin
+- all_JA_game_v2.csv: this is the final version containing path,transcript, speaker_id (not number). Game are wutheringWaves_Dataset_new, honkai_starail and genshin and Gal_game.
+- JA_phoneme.txt: this is the final version containing path,phoneme, speaker_id_name ( number). This file is maken from all_JA_game_v2.csv
+- EN_phoneme.txt: this is the final version containing path,phoneme, speaker_id_name ( number). This file is maken from all_EN_game_v2.csv
+
+# 3. Directory Structure and Dataset Repositories
+- The root path start from Data_Speech and the path given in all_JA_game_v2, all_EN_game_v2.txt start from audio_path
+- There are 2 responsity
+  + Path 1: https://huggingface.co/datasets/NoWaifuNoLaifu/TTS_dataset
+ ![alt text](image.png)
+  + Path 2: https://huggingface.co/datasets/anpham2/genshin_impact
+ ![alt text](image-1.png)
+
+- Please construct data as follow:
+
+Please pay attention to audio_2s.zip, you first need to extract gal_game_JA.zip first, then move audio_2s.zip into folder gal_game_JA and extract audio_2s.zip.
+
+All the zip file including wutheringWaves_Dataset_new.zip, genshin_japanese.zip, genshin_english.zip, honkai_starail.zip and gal_game_JA.zip just need to be added at right place then etract, nothing more.
+
+```
+audio_path/
+|____wutheringWaves_Dataset_new.zip
+|       
+|____genshin/
+|     |____genshin_japanese.zip
+|     |____genshin_english.zip
+|         
+|
+|____honkai_starail.zip
+|        
+|____gal_game_JA.zip
+     |____audio_2s.zip
+```