Update voice_create.py

voice_create.py

@@ -0,0 +1,64 @@
+import gradio as gr
+import pandas as pd
+import zipfile
+import base64
+import os
+import requests
+
+def text_to_speech(input_file):
+    # APIキーを直接コードに埋め込む（実際の運用では推奨されません）
+    api_key = 'AIzaSyAEzK5_n6zKTimD9yoXS-C8O0xN_4LaVBQ'  # ここを実際のAPIキーに置き換えてください
+    data = pd.read_csv(input_file)
+    zip_path = 'output_audio_files.zip'
+
+    with zipfile.ZipFile(zip_path, 'w') as z:
+        for idx, row in data.iterrows():
+            parts = row['script'].split("\n")
+            ssml_parts = []
+            # 交互に発言するAとBの内容を順に処理
+            for i, part in enumerate(parts):
+                if part.startswith("A:"):
+                    voice_name = row["voiceA"]
+                elif part.startswith("B:"):
+                    voice_name = row["voiceB"]
+                else:
+                    continue  # A:またはB:で始まらない行は無視
+                text = part.split(":", 1)[1].strip()
+                ssml_parts.append(f'<voice name="{voice_name}"><prosody speed="{row["speed"]}"><p>{text}</p></prosody></voice>')
+                print("読み上げ速度", row["speed"])
+
+            ssml = '<speak>' + ''.join(ssml_parts)
+            if pd.notna(row.get('question')) and row['question'] != '':
+                ssml += f'<break time="1s"/><voice name="{row["voiceQuestion"]}"><p>Question</p><break time="1s"/><p>{row["question"]}</p></voice>'
+                # Choices部分の追加
+            if pd.notna(row.get('choices')) and row['choices'] != '':
+                choices_list = row['choices'].split('/')
+                choices_ssml = '<break time="1s"/>'.join(choices_list)
+                ssml += f'<break time="1s"/><voice name="{row["voiceB"]}"><p>{choices_ssml}</p></voice>'
+            ssml += '</speak>'
+
+            # APIリクエスト用のボディ
+            body = {
+                "input": {"ssml": ssml},
+                "voice": {"languageCode": "en-US"},  # 基本的な言語設定（必要に応じて行ごとに変更可能）
+                "audioConfig": {"audioEncoding": "MP3"}
+            }
+            headers = {
+                "X-Goog-Api-Key": api_key,
+                "Content-Type": "application/json"
+            }
+            url = "https://texttospeech.googleapis.com/v1/text:synthesize"
+            response = requests.post(url, headers=headers, json=body)
+            response_data = response.json()
+
+            # 音声コンテンツの取得とファイル保存
+            if 'audioContent' in response_data:
+                audio_content = base64.b64decode(response_data['audioContent'])
+                file_name = f"{row['id']}.mp3"
+                with open(file_name, "wb") as out:
+                    out.write(audio_content)
+                
+                z.write(file_name)
+                os.remove(file_name)
+
+    return zip_path