Upgrade openai to 1.8.0

Carleslc · Jan 20, 2024 · 181f7c2 · 181f7c2
1 parent f13f7bc
commit 181f7c2
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 12 deletions.
diff --git a/AudioToText.ipynb b/AudioToText.ipynb
@@ -65,7 +65,7 @@
         "  NO_ROOT_WARNING = '|& grep -v \\\"WARNING: Running pip as the \\'root\\' user\"' # running in Colab\n",
         "\n",
         "  !pip install --no-warn-script-location --user --upgrade pip {NO_ROOT_WARNING}\n",
-        "  !pip install --root-user-action=ignore git+https://github.com/openai/whisper.git@v20231117 openai==0.28 numpy scipy deepl pydub cohere ffmpeg-python torch==2.1.0 tensorflow-probability==0.23.0 typing-extensions==4.9.0"
+        "  !pip install --root-user-action=ignore git+https://github.com/openai/whisper.git@v20231117 openai==1.8.0 numpy scipy deepl pydub cohere ffmpeg-python torch==2.1.0 tensorflow-probability==0.23.0 typing-extensions==4.9.0"
       ]
     },
     {
@@ -308,10 +308,10 @@
         "\n",
         "import torch\n",
         "\n",
-        "import openai\n",
-        "\n",
         "import math\n",
         "\n",
+        "from openai import OpenAI\n",
+        "\n",
         "# select task\n",
         "\n",
         "task = \"Transcribe\" #@param [\"Transcribe\", \"Translate to English\"]\n",
@@ -422,12 +422,14 @@
         "}\n",
         "\n",
         "if api_key:\n",
-        "  openai.api_key = api_key\n",
+        "  api_client = OpenAI(api_key=api_key)\n",
         "\n",
         "  api_supported_formats = ['mp3', 'mp4', 'mpeg', 'mpga', 'm4a', 'wav', 'webm']\n",
         "  api_max_bytes = 25 * 1024 * 1024 # 25 MB\n",
         "\n",
-        "  api_transcribe = getattr(openai.Audio, task)\n",
+        "  api_transcribe = api_client.audio.transcriptions if task == 'transcribe' else api_client.audio.translations\n",
+        "  api_transcribe = api_transcribe.create\n",
+        "  \n",
         "  api_model = 'whisper-1' # large-v2\n",
         "\n",
         "  # https://platform.openai.com/docs/api-reference/audio?lang=python\n",
@@ -578,7 +580,8 @@
         "    for api_audio_chunk_path in api_audio_chunks:\n",
         "      ## API request\n",
         "      with open(api_audio_chunk_path, 'rb') as api_audio_file:\n",
-        "        api_result = api_transcribe(api_model, api_audio_file, **api_options)\n",
+        "        api_result = api_transcribe(model=api_model, file=api_audio_file, **api_options)\n",
+        "        api_result = api_result.model_dump() # to dict\n",
         "      \n",
         "      api_segments = api_result['segments']\n",
         "      \n",

diff --git a/audiototext.py b/audiototext.py
@@ -64,7 +64,7 @@
 
 if not args.skip_install:
   os.system("pip install --user --upgrade pip")
-  os.system("pip install git+https://github.com/openai/whisper.git@v20231117 openai==0.28 numpy scipy deepl pydub cohere ffmpeg-python torch==2.1.0 tensorflow-probability==0.23.0 typing-extensions==4.9.0")
+  os.system("pip install git+https://github.com/openai/whisper.git@v20231117 openai==1.8.0 numpy scipy deepl pydub cohere ffmpeg-python torch==2.1.0 tensorflow-probability==0.23.0 typing-extensions==4.9.0")
   print()
 
 """## [Step 2] 📁 Upload your audio files to this folder
@@ -110,10 +110,10 @@
 
 import torch
 
-import openai
-
 import math
 
+from openai import OpenAI
+
 # select task
 
 task = args.task
@@ -208,12 +208,14 @@
 }
 
 if args.api_key:
-  openai.api_key = args.api_key
+  api_client = OpenAI(api_key=args.api_key)
 
   api_supported_formats = ['mp3', 'mp4', 'mpeg', 'mpga', 'm4a', 'wav', 'webm']
   api_max_bytes = 25 * 1024 * 1024 # 25 MB
 
-  api_transcribe = getattr(openai.Audio, task)
+  api_transcribe = api_client.audio.transcriptions if task == 'transcribe' else api_client.audio.translations
+  api_transcribe = api_transcribe.create
+
   api_model = 'whisper-1' # large-v2
 
   # https://platform.openai.com/docs/api-reference/audio?lang=python
@@ -364,7 +366,8 @@ def raw_split(big_chunk):
     for api_audio_chunk_path in api_audio_chunks:
       ## API request
       with open(api_audio_chunk_path, 'rb') as api_audio_file:
-        api_result = api_transcribe(api_model, api_audio_file, **api_options)
+        api_result = api_transcribe(model=api_model, file=api_audio_file, **api_options)
+        api_result = api_result.model_dump() # to dict
 
       api_segments = api_result['segments']