new

app.py

@@ -0,0 +1,189 @@
+# -*- coding: utf-8 -*-
+"""【3種用】TTS.ipynb
+
+Automatically generated by Colaboratory.
+
+Original file is located at
+    https://colab.research.google.com/drive/1Qybq46gufuZ0ykmXzSsc5EGfEN5Ucm3Z
+"""
+
+!pip install gradio==4.0.2
+!pip install openai
+
+import gradio as gr
+from openai import OpenAI
+import os
+from datetime import datetime
+from zoneinfo import ZoneInfo
+
+auto_play_bl = {'ON': True, 'OFF': False}
+voice_list = ["alloy", "echo", "fable", "onyx", "nova", "shimmer"]
+
+def set_state(state, openai_key, voice, auto_play, speed):
+
+    state["openai_key"] = openai_key
+    state["voice"] = voice
+    state["auto_play"] = auto_play_bl[auto_play]
+    state["speed"] = speed
+
+    return state
+
+def create_voice(state, text, file):
+
+    err_msg = ""
+
+    # セッション情報取得
+    client = state["client"]
+    user_id = state["user_id"]
+    voice = state["voice"]
+    speed = state["speed"]
+
+    # OpenAIキーチェック
+    if state["openai_key"] == "":
+
+        err_msg = "OpenAIキーを入力してください。（設定タブ）"
+
+        return None, err_msg
+
+    # ファイル入力チェック
+    if text.strip() == "" and file is None:
+
+        err_msg = "画面から文章を入力するか、テキストファイルをアップして下さい。"
+
+        return None, err_msg
+
+    if client is None:
+
+        os.environ["OPENAI_API_KEY"] = state["openai_key"]
+
+        # クライアント作成
+        client = OpenAI()
+
+        # client作成後は消す
+        os.environ["OPENAI_API_KEY"] = ""
+
+        state["client"] = client
+
+    if user_id == "":
+
+        # IDとして現在時刻をセット
+        dt = datetime.now(ZoneInfo("Asia/Tokyo"))
+        user_id = dt.strftime("%Y%m%d%H%M%S")
+
+        # ユーザIDでフォルダ作成
+        os.makedirs(user_id, exist_ok=True)
+
+        state["user_id"] = user_id
+
+    if file:
+
+        with open(file, 'r') as f:
+
+            input_text = f.read()
+
+    else:
+
+        input_text = text
+
+    # ファイル名は現在時刻
+    dt = datetime.now(ZoneInfo("Asia/Tokyo"))
+    file_name = dt.strftime("%Y%m%d%H%M%S") + ".mp3"
+    file_path = user_id + "/" + file_name
+
+    # 音声にする
+    result = request_tts(client, voice, speed, file_path, input_text)
+
+    if result != "":
+        err_msg = result
+        file_path = None
+
+    return file_path, err_msg
+
+def request_tts(client, voice , speed, file_path, text):
+    """ テキストを音声にする """
+
+    err_msg = ""
+
+    try:
+
+      response = client.audio.speech.create(
+          model= "tts-1",   # "tts-1-hd",
+          voice=voice,
+          input=text,
+          speed=speed
+      )
+
+      # 音声ファイルに出力
+      response.stream_to_file(file_path)
+
+    except Exception as e:
+        err_msg = "音声作成中にエラーが発生しました。"
+        print(e)
+    finally:
+        return err_msg
+
+with gr.Blocks() as demo:
+
+    title = "<h2>Text to Speechデモアプリ</h2>"
+    message = "<h3>最初に[設定]タブからOpenAIキーを入力してください。"
+    message += "</h3>"
+
+    gr.Markdown(title + message)
+
+    state = gr.State({
+        "openai_key" : "",
+        "client" : None,
+        "user_id" : "",
+        "auto_play" : True,
+        "speed" : 0.8,
+        "voice" : "nova"
+    })
+
+    with gr.Tab("音声にする") as main_tab:
+
+        text = gr.Textbox(label="音声にするテキスト", lines=3, interactive = True)
+        file = gr.File(label="入力ファイル", type="filepath",file_types=[".txt"], interactive = True)
+
+        with gr.Row():
+            btn = gr.Button("音声にする")
+            clear = gr.ClearButton([text, file], value="クリア")
+
+        sys_msg = gr.Text(label="システムメッセージ", interactive = False)
+        voice = gr.Audio(label="出力音声", type="filepath", interactive = False, autoplay = True)
+
+        btn.click(create_voice, [state, text, file], [voice, sys_msg])
+
+    with gr.Tab("設定") as set_tab:
+        openai_key = gr.Textbox(label="OpenAI API Key", interactive = True)
+        voice = gr.Dropdown(choices=voice_list, value = "nova", label="Voice", interactive = True)
+        auto_play = gr.Dropdown(choices=["ON", "OFF"], value = "ON", label="Auto Play", interactive = True)
+        speed = gr.Slider(0, 1, value=0.8, label="Speed", info="1に近づけるほど読むスピードが速くなります。", interactive = True)
+
+        # 設定変更時
+        main_tab.select(set_state, [state, openai_key, voice, auto_play, speed], state)
+
+    with gr.Tab("声サンプル") as voice_chk:
+
+        gr.Markdown("<h3>Text to speechの声のサンプルです。（速度は0.8です）</h3>")
+
+        with gr.Row():
+            btn_alloy = gr.Button(value="alloy")
+            btn_echo = gr.Button(value="echo")
+            btn_fable = gr.Button(value="fable")
+
+        with gr.Row():
+            btn_onyx = gr.Button(value="onyx")
+            btn_nova = gr.Button(value="nova")
+            btn_shimmer = gr.Button(value="shimmer")
+
+        sample_voice=gr.Audio(type="filepath", interactive = False, autoplay = True)
+
+        btn_alloy.click(lambda:"voice_sample/alloy.mp3", None, sample_voice)
+        btn_echo.click(lambda:"voice_sample/echo.mp3", None, sample_voice)
+        btn_fable.click(lambda:"voice_sample/fable.mp3", None, sample_voice)
+        btn_onyx.click(lambda:"voice_sample/onyx.mp3", None, sample_voice)
+        btn_nova.click(lambda:"voice_sample/nova.mp3", None, sample_voice)
+        btn_shimmer.click(lambda:"voice_sample/shimmer.mp3", None, sample_voice)
+
+demo.queue()
+demo.launch(debug=True)

requirements.txt

@@ -0,0 +1,2 @@
+# gradio==4.19.0
+openai==1.12.0