[WIP/TEST] Model : モデルの保存位置の変更

- speakerの文字起こし処理のバグを修正

.gitignore

@@ -6,7 +6,7 @@ VRCT.spec
 *.pyc
 logs/
 .venv/
-weight/
+weights/
 .vscode
 error.log
 *.exe

main.py

@@ -8,14 +8,13 @@ if __name__ == "__main__":
         splash.showSplash()
 
         from config import config
-        from models.translation.utils import downloadCTranslate2Weight
+        from models.translation.translation_utils import downloadCTranslate2Weight
         if config.USE_TRANSLATION_FEATURE is True:
             downloadCTranslate2Weight(config.PATH_LOCAL, config.CTRANSLATE2_WEIGHT_TYPE, splash.updateDownloadProgress)
-        splash.toProgress(0)
 
+        from models.transcription.transcription_whisper import downloadWhisperWeight
         # whisperのダウンロードの説明に変更する必要あり
         if config.USE_WHISPER_FEATURE is True:
-            from models.transcription.transcription_whisper import downloadWhisperWeight
             downloadWhisperWeight(config.PATH_LOCAL, config.WHISPER_WEIGHT_TYPE, splash.updateDownloadProgress)
         splash.toProgress(0)
 

model.py

@@ -23,7 +23,7 @@ from models.transcription.transcription_transcriber import AudioTranscriber
 from models.xsoverlay.notification import xsoverlayForVRCT
 from models.translation.translation_languages import translation_lang
 from models.transcription.transcription_languages import transcription_lang
-from models.translation.utils import checkCTranslate2Weight
+from models.translation.translation_utils import checkCTranslate2Weight
 from config import config
 
 class threadFnc(Thread):
@@ -424,7 +424,7 @@ class Model:
             root=config.PATH_LOCAL,
         )
         def sendSpeakerTranscript():
-            speaker_transcriber.transcribeAudioQueue(speaker_audio_queue, config.TARGET_LANGUAGE, config.TARGET_COUNTRY)
+            speaker_transcriber.transcribeAudioQueue(config.SELECTED_RECOGNIZER, speaker_audio_queue, config.TARGET_LANGUAGE, config.TARGET_COUNTRY)
             message = speaker_transcriber.getTranscript()
             try:
                 fnc(message)

models/transcription/transcription_whisper.py

@@ -60,10 +60,9 @@ def checkWhisperWeight(path):
     return result
 
 def downloadWhisperWeight(root, weight_type, callbackFunc):
-    path = os_path.join(root, "weight", "whisper", weight_type)
+    path = os_path.join(root, "weights", "whisper", weight_type)
     os_makedirs(path, exist_ok=True)
     if checkWhisperWeight(path) is True:
-        print("weight_type:", weight_type, checkWhisperWeight(path))
         return
 
     for filename in _FILENAMES:
@@ -72,10 +71,8 @@ def downloadWhisperWeight(root, weight_type, callbackFunc):
         url = huggingface_hub.hf_hub_url(_MODELS[weight_type], filename)
         downloadFile(url, file_path, func=callbackFunc)
 
-    print("weight_type:", weight_type, checkWhisperWeight(path))
-
 def getWhisperModel(root, weight_type):
-    path = os_path.join(root, "weight", "whisper", weight_type)
+    path = os_path.join(root, "weights", "whisper", weight_type)
     return WhisperModel(
         path,
         device="cpu",

models/translation/translation_translator.py

@@ -2,7 +2,7 @@ import os
 from deepl import Translator as deepl_Translator
 from translators import translate_text as other_web_Translator
 from .translation_languages import translation_lang
-from .utils import ctranslate2_weights
+from .translation_utils import ctranslate2_weights
 
 import ctranslate2
 import transformers
@@ -27,8 +27,8 @@ class Translator():
     def changeCTranslate2Model(self, path, model_type):
         directory_name = ctranslate2_weights[model_type]["directory_name"]
         tokenizer = ctranslate2_weights[model_type]["tokenizer"]
-        weight_path = os.path.join(path, "weight", directory_name)
+        weight_path = os.path.join(path, "weights", "ctranslate2", directory_name)
-        tokenizer_path = os.path.join(path, "weight", directory_name, "tokenizer")
+        tokenizer_path = os.path.join(path, "weights", "ctranslate2", directory_name, "tokenizer")
         self.ctranslate2_translator = ctranslate2.Translator(
             weight_path,
             device="cpu",
@@ -41,7 +41,7 @@ class Translator():
             self.ctranslate2_tokenizer = transformers.AutoTokenizer.from_pretrained(tokenizer, cache_dir=tokenizer_path)
         except Exception as e:
             print("Error: changeCTranslate2Model()", e)
-            tokenizer_path = os.path.join("./weight", directory_name, "tokenizer")
+            tokenizer_path = os.path.join("./weights", "ctranslate2", directory_name, "tokenizer")
             self.ctranslate2_tokenizer = transformers.AutoTokenizer.from_pretrained(tokenizer, cache_dir=tokenizer_path)
 
     @staticmethod

models/translation/translation_utils.py

@@ -39,36 +39,36 @@ def calculate_file_hash(file_path, block_size=65536):
     return hash_object.hexdigest()
 
 def checkCTranslate2Weight(path, weight_type="Small"):
-    directory_name = 'weight'
-    current_directory = path
     weight_directory_name = ctranslate2_weights[weight_type]["directory_name"]
     hash_data = ctranslate2_weights[weight_type]["hash"]
-    files = ["model.bin", "sentencepiece.model", "shared_vocabulary.txt"]
+    files = [
+        "model.bin",
+        "sentencepiece.model",
+        "shared_vocabulary.txt"
+    ]
 
     # check already downloaded
     already_downloaded = False
-    if all(os_path.exists(os_path.join(current_directory, directory_name, weight_directory_name, file)) for file in files):
+    if all(os_path.exists(os_path.join(path, weight_directory_name, file)) for file in files):
         # check hash
         for file in files:
             original_hash = hash_data[file]
-            current_hash = calculate_file_hash(os_path.join(current_directory, directory_name, weight_directory_name, file))
+            current_hash = calculate_file_hash(os_path.join(path, weight_directory_name, file))
             if original_hash != current_hash:
                 break
         already_downloaded = True
     return already_downloaded
 
-def downloadCTranslate2Weight(path, weight_type="Small", func=None):
+def downloadCTranslate2Weight(root, weight_type="Small", func=None):
     url = ctranslate2_weights[weight_type]["url"]
-    filename = 'weight.zip'
+    filename = "weight.zip"
-    directory_name = 'weight'
+    path = os_path.join(root, "weights", "ctranslate2")
-    current_directory = path
+    os_makedirs(path, exist_ok=True)
 
     if checkCTranslate2Weight(path, weight_type):
         return
 
     try:
-        os_makedirs(os_path.join(current_directory, directory_name), exist_ok=True)
-        print(os_path.join(current_directory, directory_name))
         with tempfile.TemporaryDirectory() as tmp_path:
             res = requests_get(url, stream=True)
             file_size = int(res.headers.get('content-length', 0))
@@ -81,6 +81,6 @@ def downloadCTranslate2Weight(path, weight_type="Small", func=None):
                         func(total_chunk/file_size)
 
             with ZipFile(os_path.join(tmp_path, filename)) as zf:
-                zf.extractall(os_path.join(current_directory, directory_name))
+                zf.extractall(path)
     except Exception as e:
             print("error:downloadCTranslate2Weight()", e)