Sample Code to Use Azure OpenAI Service's Whisper API from Python

This is a simple sample code that demonstrates how to use the Azure OpenAI Service's Whisper API to transcribe audio files using Python.

Installation and Setup

1. Ensure you have Python installed on your machine.
2. Clone the repository and navigate to the project folder.
3. Install the required dependencies by running pip install -r requirements.txt.
4. Create a .env file in the project directory and provide the necessary environment variables as mentioned below.

Environment Variables

The following environment variables need to be set in the .env file:

• OPENAI_API_TYPE: The type of API for the Azure OpenAI Service. Choose one of the supported API types: 'azure', 'azure_ad', 'open_ai'.
• OPENAI_API_HOST: The API host endpoint for the Azure OpenAI Service.
• OPENAI_API_KEY: The API key for the Azure OpenAI Service.
• OPENAI_API_VERSION: The version of the Azure OpenAI Service API.
• LANGUAGE: The language parameter for the Azure OpenAI Service. Must be specified in ISO-639-1 format.
• AZURE_DEPLOYMENT_ID: The deployment ID for the Azure AI Studio.

Code Explanation

1. First, the necessary libraries are imported: openai, os, join and dirname from os.path, and load_dotenv from dotenv.
2. The .env file is loaded to get the environment variables.
3. The parameters for the Azure OpenAI Service whisper are set based on the values read from the .env file.
4. The parameter values are confirmed by printing them.
5. The model engine name is specified as whisper-1.
6. An audio file is opened for transcription. Please note that the maou_14_shining_star.m4a file used in this code is copyrighted and can only be used as per the rules specified in the provided link.
7. The transcribe method of the Audio class from openai is executed with the specified parameters.
8. The resulting transcription text is printed.

Please make sure to replace the placeholder values in the .env file with your own valid API credentials and deployment information before running the code.

Requirements

Make sure to add the following Python packages to your requirements.txt file:

openai >= 1.0.0
python-dotenv == 1.0.1

Troubleshoot

If the REST API throws an exception and you get the following error, please refer to the instructions on how to deal with it and try to fix it.

• The API type provided in invalid. Please select one of the supported API types: 'azure', 'azure_ad', 'open_ai' ->Confirm the OPENAI_API_TYPE environment valiables.

• Access denied due to invalid subscription key or wrong API endpoint. Make sure to provide a valid key for an active subscription and use a correct regional API endpoint for your resource. ->Confirm the OPENAI_API_HOST environment valiables.

• The API deployment for this resource does not exist. If you created the deployment within the last 5 minutes, please wait a moment and try again. ->Confirm the OPENAI_API_KEY environment valiables.

• Access denied due to invalid subscription key or wrong API endpoint. Make sure to provide a valid key for an active subscription and use a correct regional API endpoint for your resource. ->Confirm the OPENAI_API_KEY environment valiables.

• Resource not found error if you make a mistake ->Confirm the OPENAI_API_VERSION environment valiables.

• Invalid language 'japanese'. Language parameter must be specified in ISO-639-1 format. ->Confirm the LANGUAGE environment valiables. and, reffer to https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes

• The API deployment for this resource does not exist. If you created the deployment within the last 5 minutes, please wait a moment and try again. ->Confirm the AZURE_DEPLOYMENT_ID environment valiables.

• Resource not found error ->I restarted Powershell running the Python commands and it fixed it. I don't know why, but I think it was probably an environment variable that was no longer being taken.

Author

• potofo

Revision

• 2023/10/08 01-00 @potofo: Initial Creation.
• 2024/06/05 01-01 @cnkang: Initial Creation. Updated openai package version in requirements.txt to openai >=1.0. Adjusted code to align with changes in the OpenAI API, ensuring compatibility with the new version. Tested all functionalities to ensure that existing features operate smoothly with the upgraded library.
• 2024/06/05 01-01 @potofo: Rename requirement.txt to requirements.txt

License

This code is licensed under the MIT License.