init

Upload 3 files.

.gitattributes

@@ -53,3 +53,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.jpg filter=lfs diff=lfs merge=lfs -text
 *.jpeg filter=lfs diff=lfs merge=lfs -text
 *.webp filter=lfs diff=lfs merge=lfs -text
+PreprocessSortPath.txt filter=lfs diff=lfs merge=lfs -text
+SortUrl_01.txt filter=lfs diff=lfs merge=lfs -text

PreprocessSortPath.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28fd4cf31b613f7eaea90a955c5512488e53854eff4dba2d1833a3e1a8e8f9d0
+size 28305148

SortUrl_01.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:668135191559e49d2821654d42658dd64d43763e1e992f4e10bbfad73b5e2035
+size 25636366

download_files_and_create_list.py

@@ -0,0 +1,82 @@
+import os, time, requests
+from pydub import AudioSegment
+
+preprocess_path = "D:/AI-DeepLearning/_Dataset/SortUrl_01.txt"
+preprocess_output_path = "D:/AI-DeepLearning/_Dataset/PreprocessSortPathUrl.txt"
+output_path = "D:/AI-DeepLearning/_Dataset/ViTS/BanGDream"
+sub_path = "VoiceData"
+
+# The path of audio when your training
+write_in_path = "D:/AI-DeepLearning/Bert-VITS2-Integration-package/Data/BanGDream/custom_character_voice"
+
+
+def download_mp3_and_convert_to_wav(preprocess_path, output_path, sub_path):
+    data_list = []
+
+    # Path to save the download audio
+    voice_output_path = os.path.join(output_path, sub_path)
+    if not os.path.exists(voice_output_path):
+        os.mkdir(voice_output_path)
+
+    with open(preprocess_path, "r", encoding="utf-8") as f:
+        for line in f.readlines():
+            data_list.append(line.strip("\n"))
+
+        f.close()
+
+    # If you want overwrite please exchange 'a+' to 'w'
+    # , and change line56 if-loop to avoid passing existfile write.
+    with open(preprocess_output_path, "a+", encoding="utf-8") as f:
+        count = 0
+        for line in data_list:
+            url = line.split("|")[0]
+            speaker = line.split("|")[1]
+            script = line.split("|")[2]
+
+            # Check is it an url
+            if not 'https' in url:
+                print(f'{url} is not a url.')
+                count += 1
+                print(f'{count}/{len(data_list)}')
+                continue
+
+            # Find speaker
+            speaker_path = os.path.join(voice_output_path, speaker)
+            
+            if not os.path.exists(speaker_path):
+                os.mkdir(speaker_path)
+
+
+            # Mp3
+            mp3_file = url.split("/")[-1]
+            mp3_path = os.path.join(speaker_path, mp3_file)
+            save_wav_path = mp3_path.replace(".mp3", ".wav")
+
+            download = requests.get(url)
+            if not os.path.exists(mp3_path) and not os.path.exists(save_wav_path):
+                if download.status_code == 200:
+                    with open(mp3_path, "wb") as voice:
+                        voice.write(download.content)
+                        voice.close()
+
+
+                #　Turn wav and delete mp3
+                audio = AudioSegment.from_mp3(mp3_path)
+                audio = audio.set_frame_rate(44100).set_channels(1)
+                audio.export(save_wav_path, format="wav")
+                os.remove(mp3_path)
+
+                wav_file = mp3_file.replace(".mp3", ".wav")
+                write_in_wav_path = os.path.join(write_in_path, speaker, wav_file)
+
+                f.write(f'{write_in_wav_path}|{speaker}|JP|{script}\n')
+            else:
+                mp3_file = mp3_file.replace('.mp3', '')
+                print(f'{mp3_file} already exists.')
+
+            count += 1
+            print(f'{count}/{len(data_list)}')
+    
+
+if __name__ == "__main__":
+    download_mp3_and_convert_to_wav(preprocess_path=preprocess_path, output_path=output_path, sub_path=sub_path)