description: "An example Media Workflow with Dolby.io presented at the 2021 Azure Serverless Conf"
languages:

• python
  products:
• azure-functions
• azure-storage
• azure-cognitive-services

Media Pipeline

This sample highlights how to use Azure functions and Dolby.io to create a media processing pipeline that transcribes and enhances input media.

Getting Started

Prerequisites

• Install Python 3.8+
• Azure Storage Account
• Azure Speech-to-text service Account
• Dolby.io Account, sign up for free
• ngrok account (only for local testing), sign up for free

Params.json

• "cog_api_key": Azure Cognitive Services Speech-to-text API Key,

• "dolbyio_api_key": Dolby.io media API Key,

• "azure_api_key" : Azure storage account API key,

• "AZURE_ACC_NAME" : Azure storage account name,

• "AZURE_CONTAINER" : Azure storage container name,

• "AZURE_BLOB_OUTPUT_SUFFIX" : the output suffix added to the enhanced version of the input media,

• "SCORE_THRESHOLD" : the score threshold that decides if the audio quality of the input media is too low,

• "transcription_url" : the Azure cognitive services input URL for transcription, dependent on region.

• "tunneling_url" : The location of your Serverless HTTP trigger, or your HTTP tunnel (if you run locally)

Steps

• Start by cloning this repo to your local machine.

Local with VS Code (Recommended for development)

• Follow the guide here to get setup with VS Code, Azure functions, and Python.
• Once set up you can clone the project and open it in VS Code.
• Create a HTTP tunnel with ngrok
• • "ngrok" is a tunneling tool that allows you to port forward to local host.
• • In my case my local function was deployed on 7071, so I initalized ngrok to port forward on LocalHost:7071
• Adjust the "params.json" file with your HTTP ngrok Tunnel and the other appropriate params such as your blob name and API keys, then save the file.
• Now you can press F5 to run the project.
• Once the code is running, navigate to http://localhost:"YOUR_SERVER"/api/MediaProcessingWorkflow?input_file="YOUR_INPUT_FILE" in your browser.
• • This will trigger the function to run locally, if all params are valid you can check your storage account for transcriptions or enhanced media.

Server with VS Code (Recommended for deployment)

• Follow the guide here to get setup with VS Code, Azure functions, and Python.
• Once set up you can clone the project and open it in VS Code.
• Adjust the "params.json" file.
• • In this case the tunneling URL will be https://"YOUR_FUNCTION_APP_NAME".azurewebsites.net/api/"YOUR_FUNCTION_NAME"
• • For example our tunneling URL is https://DolbyioMediaWorkflowTest.azurewebsites.net/api/MediaProcessingWorkflow
• Next deploy function app
• Once successfully deployed to the Azure Server you can test your function by navigating to:
• • https://"YOUR_FUNCTION_APP_NAME".azurewebsites.net/api/"YOUR_FUNCTION_NAME"?input_file="YOUR_INPUT_FILE"

References

• Use Dolby.io to inspect, correct, and perfect your audio.
• Azure Functions.
• Reach out for clarification @BradenRiggs1 on Twitter.