Update app.py

app.py

@@ -1,44 +1,82 @@
 from flask import Flask, request, jsonify, send_from_directory
 import base64
 import os
+import shutil
+import numpy as np
+from pyannote.audio import Model, Inference
+from pydub import AudioSegment
+
+os.environ["HUGGINGFACE_HUB_TOKEN"] = os.environ.get("HF")  # トークンを適切に設定
+
+# 事前学習済みモデルの読み込み
+model = Model.from_pretrained("pyannote/embedding")
+inference = Inference(model)
+
+def cosine_similarity(vec1, vec2):
+    vec1 = vec1 / np.linalg.norm(vec1)
+    vec2 = vec2 / np.linalg.norm(vec2)
+    return np.dot(vec1, vec2)
+
+def segment_audio(path, target_path='./setup_voice', seg_duration=1.0):
+    """音声を指定秒数ごとに分割する"""
+    os.makedirs(target_path, exist_ok=True)
+    base_sound = AudioSegment.from_file(path)
+    duration_ms = len(base_sound)
+    seg_duration_ms = int(seg_duration * 1000)
+    
+    for i, start in enumerate(range(0, duration_ms, seg_duration_ms)):
+        end = min(start + seg_duration_ms, duration_ms)
+        segment = base_sound[start:end]
+        segment.export(os.path.join(target_path, f'{i}.wav'), format="wav")
+    
+    return target_path, duration_ms
+
+def calculate_similarity(path1, path2):
+    embedding1 = inference(path1)
+    embedding2 = inference(path2)
+    return float(cosine_similarity(embedding1.data.flatten(), embedding2.data.flatten()))
+
+def process_audio(reference_path, input_path, output_folder='/tmp/data/matched_segments', seg_duration=1.0, threshold=0.5):
+    os.makedirs(output_folder, exist_ok=True)
+    base_path, total_duration_ms = segment_audio(input_path, seg_duration=seg_duration)
+    
+    matched_time_ms = 0
+    for file in sorted(os.listdir(base_path)):
+        segment_file = os.path.join(base_path, file)
+        similarity = calculate_similarity(segment_file, reference_path)
+        if similarity > threshold:
+            shutil.copy(segment_file, output_folder)
+            matched_time_ms += len(AudioSegment.from_file(segment_file))
+    
+    unmatched_time_ms = total_duration_ms - matched_time_ms
+    return matched_time_ms, unmatched_time_ms
 
 app = Flask(__name__)
 
 @app.route('/')
 def index():
-    return send_from_directory(".", "index.html")
+    return send_from_directory('.', 'index.html')
 
 @app.route('/upload_audio', methods=['POST'])
 def upload_audio():
     try:
         data = request.get_json()
-        if not data:
-            return jsonify({"error": "JSONが送信されていません"}), 400
-
-        audio_data = data.get('audio_data')
-        if not audio_data:
-            return jsonify({"error": "音声データが送信されていません"}), 400
-
-        # Base64デコード
-        try:
-            audio_binary = base64.b64decode(audio_data)
-        except Exception as decode_err:
-            return jsonify({"error": "Base64デコードに失敗しました", "details": str(decode_err)}), 400
-
-        # 書き込み用ディレクトリとして /tmp/data を使用（/tmp は書き込み可能）
-        persist_dir = "/tmp/data"
-        os.makedirs(persist_dir, exist_ok=True)
-
-        filepath = os.path.join(persist_dir, "recorded_audio.wav")
-        with open(filepath, 'wb') as f:
+        if not data or 'audio_data' not in data:
+            return jsonify({"error": "音声データがありません"}), 400
+        
+        audio_binary = base64.b64decode(data['audio_data'])
+        audio_path = "/tmp/data/recorded_audio.wav"
+        os.makedirs(os.path.dirname(audio_path), exist_ok=True)
+        with open(audio_path, 'wb') as f:
             f.write(audio_binary)
-
-        return jsonify({"message": "音声が正常に保存されました", "filepath": filepath}), 200
-
+        
+        reference_audio = './sample.wav'  # 参照音声
+        matched_time, unmatched_time = process_audio(reference_audio, audio_path, threshold=0.1)
+        rate = (matched_time / (matched_time + unmatched_time)) * 100 if (matched_time + unmatched_time) > 0 else 0
+        
+        return jsonify({"rate": rate}), 200
     except Exception as e:
-        app.logger.error("エラー: %s", str(e))
-        return jsonify({"error": "サーバー内部エラー", "details": str(e)}), 500
+        return jsonify({"error": "サーバーエラー", "details": str(e)}), 500
 
 if __name__ == '__main__':
-    port = int(os.environ.get("PORT", 7860))
-    app.run(debug=True, host="0.0.0.0", port=port)
+    app.run(debug=True, host="0.0.0.0", port=7860)