Update transcription.py

app.py

@@ -13,13 +13,13 @@ transcripter = TranscriptionMaker()
 app = Flask(__name__)
 
 users = []
-segments_dir = ""
 transcription_text=""
 harassment_keywords = [
         "バカ", "馬鹿", "アホ", "死ね", "クソ", "うざい",
         "きもい", "キモい", "ブス", "デブ", "ハゲ",
         "セクハラ", "パワハラ", "モラハラ"
     ]
+total_audio = ""
 
 # トップページ（テンプレート: index.html）
 @app.route('/')
@@ -62,10 +62,10 @@ def reset_member():
 # 書き起こし作成エンドポイント
 @app.route('/transcription',methods =['GET','POST'])
 def transcription():
-    global segments_dir
     global transcription_text
+    global total_audio
     try:
-        transcription_text = transcripter.create_transcription(segments_dir)
+        transcription_text = transcripter.create_transcription(total_audio)
         print(transcription_text)
         with open(transcription_text,'r',encoding='utf-8') as file:
             file_content = file.read()
@@ -109,7 +109,7 @@ def analyze():
 # 音声アップロード＆解析エンドポイント
 @app.route('/upload_audio', methods=['POST'])
 def upload_audio():
-    global segments_dir
+    global total_audio
     try:
         data = request.get_json()
         # name か users のいずれかが必須。どちらも無い場合はエラー
@@ -139,12 +139,14 @@ def upload_audio():
         if len(users) > 1:
             print("複数人の場合の処理")
             matched_times, segments_dir = process.process_multi_audio(reference_paths, audio_path, threshold=0.05)
+            total_audio = transcripter.merge_segments(segments_dir)
             # 各メンバーのrateを計算
             total_time = sum(matched_times)
             rates = [(time / total_time) * 100 if total_time > 0 else 0 for time in matched_times]
             return jsonify({"rates": rates}), 200
         else:
             matched_time, unmatched_time, segments_dir = process.process_audio(reference_paths[0], audio_path, threshold=0.05)
+            total_audio = transcripter.merge_segments(segments_dir)
             total_time = matched_time + unmatched_time
             rate = (matched_time / total_time) * 100 if total_time > 0 else 0
             return jsonify({"rate": rate}), 200

process.py

@@ -58,15 +58,7 @@ class AudioProcessor():
         embedding2 = self.inference(path2)
         return float(self.cosine_similarity(embedding1.data.flatten(), embedding2.data.flatten()))
     
-    def generate_random_string(self,length):
-        letters = string.ascii_letters + string.digits
-        return ''.join(random.choice(letters) for i in range(length))
-
-    def generate_filename(self,random_length):
-        random_string = self.generate_random_string(random_length)
-        current_time = datetime.now().strftime("%Y%m%d%H%M%S")
-        filename = f"{current_time}_{random_string}.wav"
-        return filename
+
     
     def process_audio(self, reference_path, input_path, output_folder='/tmp/data/matched_segments', seg_duration=1.0, threshold=0.5):
         # 出力先ディレクトリの中身をクリアする

transcription.py

@@ -1,6 +1,9 @@
 import os
 from faster_whisper import WhisperModel
 from pydub import AudioSegment
+import string
+import random
+from datetime import datetime
 
 # Matplotlibのキャッシュディレクトリを変更
 os.environ["MPLCONFIGDIR"] = "/tmp/matplotlib"
@@ -18,10 +21,9 @@ class TranscriptionMaker():
 
 
     #音声ファイルのディレクトリを受け取り、書き起こしファイルを作成する
-    def create_transcription(self,segments_directory):
+    def create_transcription(self,audio_directory):
         results = []
-        #細切れ音声をくっつける
-        audio_directory = self.merge_segments(segments_directory)
+
         #ディレクトリ内のファイルを全て取得
         if not os.path.isdir(audio_directory):
             raise ValueError(f"The specified path is not a valid directory: {audio_directory}")
@@ -53,7 +55,7 @@ class TranscriptionMaker():
         return output_file
     
     #ファイル名が連続しているならくっつける
-    def merge_segments(self,segments_dir,output_dir = "/tmp/data/merged_audio"):
+    def merge_segments(self,segments_dir,output_dir = "/tmp/data/merged_segment"):
         if not os.path.exists(output_dir):
             os.makedirs(output_dir, exist_ok=True)
 
@@ -90,7 +92,17 @@ class TranscriptionMaker():
                 segment = AudioSegment.from_file(file_path)
                 combined_audio += segment
             # 出力ファイル名を設定して保存
-            output_file = os.path.join(output_dir, f'merged_{i}.wav')
+            output_file = os.path.join(output_dir, self.generate_filename(3))
             combined_audio.export(output_file, format='wav')
 
-        return output_dir
+        return output_dir
+    
+    def generate_random_string(self,length):
+        letters = string.ascii_letters + string.digits
+        return ''.join(random.choice(letters) for i in range(length))
+
+    def generate_filename(self,random_length):
+        random_string = self.generate_random_string(random_length)
+        current_time = datetime.now().strftime("%Y%m%d%H%M%S")
+        filename = f"{current_time}_{random_string}.wav"
+        return filename