{"id":849,"date":"2023-04-13T16:50:23","date_gmt":"2023-04-13T07:50:23","guid":{"rendered":"https:\/\/bocek.co.jp\/media\/?p=849"},"modified":"2023-04-29T20:10:50","modified_gmt":"2023-04-29T11:10:50","slug":"%e3%80%90chatgpt%e3%80%91whisperapi%e3%81%a7ai%e3%81%a8%e9%9f%b3%e5%a3%b0%e4%bc%9a%e8%a9%b1%e3%81%a7%e3%81%8d%e3%82%8bpython%e3%82%a2%e3%83%97%e3%83%aa%e3%82%92%e4%bd%9c%e3%82%8b","status":"publish","type":"post","link":"https:\/\/taskhub.jp\/magazine\/ai-development\/849\/","title":{"rendered":"\u3010ChatGPT\u3011WhisperAPI\u3067AI\u3068\u97f3\u58f0\u4f1a\u8a71\u3067\u304d\u308bPython\u30a2\u30d7\u30ea\u3092\u4f5c\u308b"},"content":{"rendered":"\n<p>\u4eca\u56de\u306fChatGPT\u3068Whisper\u306eAPI\u3092\u5229\u7528\u3057\u3066\u3001AI(ChatGPT)\u3068\u97f3\u58f0\u3067\u4f1a\u8a71\u3067\u304d\u308b\u30a2\u30d7\u30ea\u3092Python\u3067\u4f5c\u6210\u3057\u3066\u307f\u307e\u3059\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u5927\u307e\u304b\u306a\u624b\u9806<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">\u2460gradio\u3067\u97f3\u58f0\u5165\u529b\u30fb\u51fa\u529b\u306e\u30a4\u30f3\u30bf\u30fc\u30d5\u30a7\u30fc\u30b9\u3092\u4f5c\u308b<\/h3>\n\n\n\n<p>gradio\u306fPython\u30d7\u30ed\u30b0\u30e9\u30e0\u306b\u304a\u3044\u3066\u3001\u6a5f\u68b0\u5b66\u7fd2\u306e\u30e2\u30c7\u30eb\u306e\u51fa\u529b\u7d50\u679c\u3092\u7c21\u5358\u306bWeb\u30a2\u30d7\u30ea\u30b1\u30fc\u30b7\u30e7\u30f3\u306b\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u308b\u30e9\u30a4\u30d6\u30e9\u30ea\u3067\u3059\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">\u2461Whisper\u3067\u767a\u8a00\u3092\u6587\u5b57\u8d77\u3053\u3057<\/h3>\n\n\n\n<p>WhisperAPI\u3092\u4f7f\u7528\u3057\u3066\u3001\u97f3\u58f0\u3092\u6587\u5b57\u8d77\u3053\u3057\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">\u2462ChatGPT\u3067\u767a\u8a00\u304b\u3089\u8fd4\u7b54\u3092\u5f97\u308b<\/h3>\n\n\n\n<p>ChatGPT\u306eAPI\u3092\u4f7f\u7528\u3057\u3066\u3001\u6587\u5b57\u8d77\u3053\u3057\u3057\u305f\u767a\u8a00\u304b\u3089\u8fd4\u7b54\u3092\u5f97\u307e\u3059\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">\u2463gTTS\u3067\u8fd4\u7b54\u3092\u97f3\u58f0\u306b\u3059\u308b<\/h3>\n\n\n\n<p>gTTS\u3092\u4f7f\u7528\u3057\u3066\u3001\u8fd4\u7b54\u3055\u308c\u305f\u6587\u7ae0\u3092\u97f3\u58f0\u306b\u5909\u63db\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<p>gTTS\u3068\u306f\u300cGoogle Text To Speach\u300d\u306e\u7565\u3067\u3059\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">\u2464Pydub\u3067\u901f\u5ea6\u8abf\u6574<\/h3>\n\n\n\n<p>Pydub\u306f\u3001Python\u3067\u97f3\u58f0\u30d5\u30a1\u30a4\u30eb\u3092\u7de8\u96c6\u3059\u308b\u305f\u3081\u306e\u30aa\u30fc\u30d7\u30f3\u30bd\u30fc\u30b9\u306e\u30e9\u30a4\u30d6\u30e9\u30ea\u3067\u3059\u3002<\/p>\n\n\n\n<p>\u81ea\u7136\u306a\u4f1a\u8a71\u306e\u30b9\u30d4\u30fc\u30c9\u306b\u306a\u308b\u3088\u3046\u306b\u8abf\u6574\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Whisper\u3068\u306f<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Whisper\u306e\u6982\u8981<\/h3>\n\n\n\n<p>\u97f3\u58f0\u30d5\u30a1\u30a4\u30eb\u304b\u3089\u975e\u5e38\u306b\u9ad8\u3044\u7cbe\u5ea6\u3067\u6587\u5b57\u8d77\u3053\u3057\u304c\u3067\u304d\u308b\u3001\u97f3\u58f0\u8a8d\u8b58AI\u306b\u306a\u3063\u3066\u3044\u307e\u3059\u3002<\/p>\n\n\n\n<p>API\u3092\u4f7f\u7528\u3057\u3066\u97f3\u58f0\u30c7\u30fc\u30bf\u3092\u9001\u308b\u3060\u3051\u3067\u6587\u5b57\u8d77\u3053\u3057\u306e\u7d50\u679c\u3092\u8fd4\u3057\u3066\u304f\u308c\u307e\u3059\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">WhisperAPI\u306e\u6599\u91d1\u4f53\u7cfb<\/h3>\n\n\n\n<p>1\u5206\uff1d0.006\u30c9\u30eb(0.8\u5186)<\/p>\n\n\n\n<p>10\u5206\u3060\u30688\u5186\u3067\u3059\u3002<\/p>\n\n\n\n<p>OpenAI\u306e\u30a2\u30ab\u30a6\u30f3\u30c8\u3092\u4f5c\u6210\u3059\u308b\u306818\u30c9\u30eb\u306e\u7121\u6599\u67a0\u304c\u3042\u308a\u307e\u3059\u3002<\/p>\n\n\n\n<p>\u500b\u4eba\u5229\u7528\u3059\u308b\u969b\u306b\u306f\u3053\u3061\u3089\u3067\u5341\u5206\u3067\u3057\u3087\u3046\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">WhisperAPI\u306e\u53d6\u5f97\u65b9\u6cd5<\/h3>\n\n\n\n<p>ChatGPT\u306eAPI\u30ad\u30fc\u3068\u540c\u69d8\u3067\u3059\u3002<\/p>\n\n\n\n<p>\u307e\u3060\u4f5c\u6210\u3055\u308c\u3066\u3044\u306a\u3044\u65b9\u306f<a href=\"https:\/\/platform.openai.com\/account\/api-keys\" data-type=\"URL\" data-id=\"https:\/\/platform.openai.com\/account\/api-keys\">\u3053\u3061\u3089<\/a>\u304b\u3089\u300cCreate new secret key\u300d\u3092\u30af\u30ea\u30c3\u30af\u3057\u3066API\u30ad\u30fc\u3092\u4f5c\u6210\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u5b9f\u969b\u306b\u5bfe\u8a71\u30a2\u30d7\u30ea\u3092\u4f5c\u3063\u3066\u307f\u308b<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">\u6e96\u5099<\/h3>\n\n\n\n<p>\u307e\u305a\u306f\u5fc5\u8981\u306a\u30e9\u30a4\u30d6\u30e9\u30ea\u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>pip install openai gradio gtts pydub<\/code><\/pre>\n\n\n\n<p>\u74b0\u5883\u306b\u3088\u3063\u3066\u306fffmpeg\u306e\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u304c\u5fc5\u8981\u3067\u3059\u3002 <\/p>\n\n\n\n<p>\u3053\u3061\u3089\u306f<code>pydub<\/code>\u304c\u30aa\u30fc\u30c7\u30a3\u30aa\u30d5\u30a1\u30a4\u30eb\u3092\u51e6\u7406\u3059\u308b\u305f\u3081\u306b\u5fc5\u8981\u306a\u3082\u306e\u3068\u306a\u3063\u3066\u3044\u307e\u3059\u3002<\/p>\n\n\n\n<p>Mac\u306e\u5834\u5408\u3001Homebrew\u3092\u4f7f\u7528\u3059\u308b\u306e\u304c\u6700\u3082\u7c21\u5358\u3067\u3059\u3002\u307e\u3060\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u3066\u3044\u306a\u3044\u5834\u5408\u306f\u3001<a href=\"https:\/\/brew.sh\/\">Homebrew\u306e\u516c\u5f0f\u30b5\u30a4\u30c8<\/a>\u304b\u3089\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u3066\u304f\u3060\u3055\u3044\u3002\u6b21\u306b\u3001\u30bf\u30fc\u30df\u30ca\u30eb\u3067\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u3066<code>ffmpeg<\/code>\u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<pre class=\"wp-block-preformatted\"><code>brew install ffmpeg<\/code><\/pre>\n\n\n\n<p>Windows\u306e\u5834\u5408\u306f\u3001<a href=\"https:\/\/ffmpeg.org\/download.html\">\u516c\u5f0f\u306eFFmpeg\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u30da\u30fc\u30b8<\/a>\u304b\u3089Windows\u30d3\u30eb\u30c9\u3092\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3057\u3066\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u3066\u304f\u3060\u3055\u3044\u3002\u307e\u305f\u3001FFmpeg\u306e\u5b9f\u884c\u30d5\u30a1\u30a4\u30eb\u304c\u542b\u307e\u308c\u308b\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\uff08<code>bin<\/code>\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\uff09\u3092\u30b7\u30b9\u30c6\u30e0\u306ePATH\u306b\u8ffd\u52a0\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002<\/p>\n\n\n\n<p>Linux\u306e\u5834\u5408\u3001\u30c7\u30a3\u30b9\u30c8\u30ea\u30d3\u30e5\u30fc\u30b7\u30e7\u30f3\u306b\u5fdc\u3058\u305f\u30d1\u30c3\u30b1\u30fc\u30b8\u30de\u30cd\u30fc\u30b8\u30e3\u3092\u4f7f\u7528\u3057\u3066<code>ffmpeg<\/code>\u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u307e\u3059\u3002\u305f\u3068\u3048\u3070\u3001Ubuntu\u3092\u4f7f\u7528\u3057\u3066\u3044\u308b\u5834\u5408\u306f\u3001\u6b21\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<pre class=\"wp-block-preformatted\"><code>sudo apt-get update\nsudo apt-get install ffmpeg<\/code><\/pre>\n\n\n\n<h3 class=\"wp-block-heading\">\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u66f8\u304f<\/h3>\n\n\n\n<p>\u3053\u306e\u3088\u3046\u306a\u30b3\u30fc\u30c9\u306b\u306a\u308a\u307e\u3057\u305f\u3002<\/p>\n\n\n\n<p>\u5b9f\u884c\u3059\u308b\u969b\u306fAPI\u30ad\u30fc\u306e\u5165\u529b\u3092\u5fd8\u308c\u306a\u3044\u3067\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p>\u5fc5\u8981\u306b\u5fdc\u3058\u3066\u97f3\u58f0\u306e\u30b9\u30d4\u30fc\u30c9\u3084\u4f1a\u8a71\u306e\u8a2d\u5b9a\u3092\u5909\u66f4\u3057\u3066\u307f\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>import openai\nimport gradio as gr\nfrom gtts import gTTS\nfrom pydub import AudioSegment\nfrom pydub.playback import play\nimport io\n\n# Whisper API\u306eAPI\u30ad\u30fc\u3092\u8a2d\u5b9a\nopenai.api_key = \"\u3053\u3053\u306bAPI\u30ad\u30fc\u3092\u5165\u308c\u3066\u304f\u3060\u3055\u3044\"\n\n# ChatGPT\u306b\u3088\u308b\u4f1a\u8a71\u306e\u95a2\u6570\ndef chat(input_text):\n    response = openai.ChatCompletion.create(\n        model=\"gpt-3.5-turbo\",\n        messages=&#91;\n                {'role': 'system', 'content': '\u3042\u306a\u305f\u306f\u79c1\u306e\u53cb\u9054\u3067\u3059\u3002\u3042\u307e\u308a\u77e5\u8b58\u3092\u6301\u3063\u3066\u3044\u306a\u3044\u3067\u3059\u3002\u8fd4\u7b54\u306f\u5fc5\u305a20\u6587\u5b57\u4ee5\u5185\u3067\u77ed\u304f\u7b54\u3048\u3066\u304f\u3060\u3055\u3044\u3002'},\n                # \u3053\u3053\u3067\u8a2d\u5b9a\uff08\u4f1a\u8a71\u76f8\u624b\u306e\u4eba\u683c\u306a\u3069\uff09\u3092\u5909\u66f4\u3067\u304d\u307e\u3059\u3002\n                {\"role\": \"user\", \"content\":input_text}\n                # \u3053\u3053\u306b\u30e6\u30fc\u30b6\u30fc\u306e\u767a\u8a00\u304c\u5165\u308a\u307e\u3059\u3002\n                ],\n                temperature=0.0,\n                # \u5275\u9020\u6027\u3092\u8abf\u6574\u3057\u307e\u3059\u30021.0\u304c\u4e0a\u9650\u3067\u3059\u3002\n\n    )\n    return response&#91;\"choices\"]&#91;0]&#91;\"message\"]&#91;\"content\"]\n\n# \u97f3\u58f0\u304b\u3089\u30c6\u30ad\u30b9\u30c8\u3078\u306e\u5909\u63db\u95a2\u6570\ndef speech_to_text(input_audio):\n    audio_file = open(input_audio, \"rb\")\n    response = openai.Audio.transcribe(\n        \"whisper-1\", audio_file\n    )\n    return response&#91;\"text\"]\n\n# \u30c6\u30ad\u30b9\u30c8\u304b\u3089\u97f3\u58f0\u3078\u306e\u5909\u63db\u95a2\u6570\ndef text_to_speech(input_text):\n    tts = gTTS(text=input_text, lang=\"ja\")\n    # \u30c6\u30ad\u30b9\u30c8\u304b\u3089\u97f3\u58f0\u306b\u5909\u63db\n    tts.save(\"sample.mp3\") \n    # \u97f3\u58f0\u30d5\u30a1\u30a4\u30eb\u306b\u4fdd\u5b58\n    sound = AudioSegment.from_mp3(\"sample.mp3\")\n    sound_speedup = sound.speedup(playback_speed=1.5)\n    # \u8aad\u307f\u4e0a\u3052\u30b9\u30d4\u30fc\u30c9\u3092\u4e0a\u3052\u308b\n    sound_speedup.export(\"sample.mp3\", format=\"mp3\")\n    return \"sample.mp3\"\n\n# \u97f3\u58f0\u4f1a\u8a71\u30a2\u30d7\u30ea\u306e\u95a2\u6570\uff08\u4eca\u307e\u3067\u306e\u95a2\u6570\u306e\u7d44\u307f\u5408\u308f\u305b\uff09\ndef voice_chat(input_audio):\n    text = speech_to_text(input_audio)\n    # \u97f3\u58f0\u3092\u30c6\u30ad\u30b9\u30c8\u306b\n    response_text = chat(text)\n    # \u30c6\u30ad\u30b9\u30c8\u304b\u3089\u8fd4\u7b54\u3092\u751f\u6210\n    response_audio = text_to_speech(response_text)\n    # \u8fd4\u7b54\u3092\u97f3\u58f0\u306b\n    return response_audio\n\n# Gradio\u30a4\u30f3\u30bf\u30fc\u30d5\u30a7\u30fc\u30b9\ngr.Interface(\n    fn=voice_chat,\n    inputs=gr.components.Audio(source=\"microphone\",type=\"filepath\"),\n    outputs=gr.components.Audio(type=\"numpy\"),\n    examples=&#91;],\n).launch()\n<\/code><\/pre>\n\n\n\n<p>\u5fc5\u8981\u306b\u5fdc\u3058\u3066ChatGPT\u306e\u5f79\u5272\u306a\u3069\u3092\u5909\u66f4\u3057\u3066\u307f\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u5b9f\u884c\u7d50\u679c<\/h2>\n\n\n\n<p>\u30b3\u30fc\u30c9\u3092\u30d5\u30a1\u30a4\u30eb\u306b\u4fdd\u5b58\u3057\u3066\u5b9f\u884c\u3059\u308b\u3068\u3001\u30bf\u30fc\u30df\u30ca\u30eb\u306bURL\u304c\u8868\u793a\u3055\u308c\u308b\u306e\u3067\u30af\u30ea\u30c3\u30af\u3057\u3066UI\u3092\u8868\u793a\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/taskhub.jp\/magazine\/wp-content\/themes\/the-thor\/img\/dummy.gif\" data-layzr=\"https:\/\/bocek.co.jp\/media\/wp-content\/uploads\/2023\/04\/image-28-1024x672.png\" alt=\"\" class=\"wp-image-859\"\/><\/figure>\n\n\n\n<p>\u300cRecord from microphone\u300d\u3092\u30af\u30ea\u30c3\u30af\u3057\u3066\u3001\u8a71\u3057\u7d42\u308f\u3063\u305f\u3089\u300cStop recording\u300d\u3092\u62bc\u3057\u3066\u304b\u3089\u9001\u4fe1\u3057\u307e\u3057\u3087\u3046\u3002<\/p>\n\n\n\n<p>\u51e6\u7406\u304c\u5b8c\u4e86\u3059\u308b\u3068\u8fd4\u7b54\u304c\u518d\u751f\u3067\u304d\u308b\u3088\u3046\u306b\u306a\u308a\u307e\u3059\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/taskhub.jp\/magazine\/wp-content\/themes\/the-thor\/img\/dummy.gif\" data-layzr=\"https:\/\/bocek.co.jp\/media\/wp-content\/uploads\/2023\/04\/image-27-1024x558.png\" alt=\"\" class=\"wp-image-858\"\/><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\">\u307e\u3068\u3081<\/h2>\n\n\n\n<p>\u4eca\u56de\u306fChatGPT\u3068Whisper\u306eAPI\u3092\u5229\u7528\u3057\u3066\u3001ChatGPT\u3068\u97f3\u58f0\u3067\u4f1a\u8a71\u3067\u304d\u308b\u30a2\u30d7\u30ea\u3092Python\u3067\u4f5c\u6210\u3057\u307e\u3057\u305f\u3002\u82f1\u4f1a\u8a71\u306e\u7df4\u7fd2\u306a\u3069\u306b\u5fdc\u7528\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u305d\u3046\u3067\u3059\u3002\u305c\u3072\u30ab\u30b9\u30bf\u30de\u30a4\u30ba\u306a\u3069\u3092\u8a66\u3057\u3066\u307f\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u4eca\u56de\u306fChatGPT\u3068Whisper\u306eAPI\u3092\u5229\u7528\u3057\u3066\u3001AI(ChatGPT)\u3068\u97f3\u58f0\u3067\u4f1a\u8a71\u3067\u304d\u308b\u30a2\u30d7\u30ea\u3092Python\u3067\u4f5c\u6210\u3057\u3066\u307f\u307e\u3059\u3002 \u5927\u307e\u304b\u306a\u624b\u9806 \u2460gradio\u3067\u97f3\u58f0\u5165\u529b\u30fb\u51fa\u529b\u306e\u30a4\u30f3\u30bf\u30fc\u30d5\u30a7\u30fc\u30b9\u3092 [&hellip;]<\/p>\n","protected":false},"author":5,"featured_media":871,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_eb_attr":""},"categories":[50],"tags":[],"_links":{"self":[{"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/posts\/849"}],"collection":[{"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/comments?post=849"}],"version-history":[{"count":3,"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/posts\/849\/revisions"}],"predecessor-version":[{"id":870,"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/posts\/849\/revisions\/870"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/media\/871"}],"wp:attachment":[{"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/media?parent=849"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/categories?post=849"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/taskhub.jp\/magazine\/wp-json\/wp\/v2\/tags?post=849"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}