Skip to content

Latest commit

 

History

History
63 lines (45 loc) · 3.01 KB

README.md

File metadata and controls

63 lines (45 loc) · 3.01 KB
Raw

Serverless Timestamper

A serverless API that uses AI to generate YouTube video timestamps for you.

Install

How to run the API

  1. In the api folder, create a file called local.settings.json
  2. Add the following to that file and make sure to populate with your OPENAI API key and Azure SignalR Service connection string.
{
  "IsEncrypted": false,
  "Values": {
    "AzureWebJobsStorage": "UseDevelopmentStorage=true",
    "FUNCTIONS_WORKER_RUNTIME": "dotnet-isolated",
    "OPENAI_API_KEY": "",
    "AzureSignalRConnectionString": ""
  },
  "Host": {
    "CORS": "*"
  }
}
  1. In the api folder use func host start command (or you can use debugger with F5) to run the API.
  2. Make a get call to the URL of the DurableFunctionsOrchestrationCSharp1_HttpStart function and provide the videoUrl parameter with the url of the video you want to generate timestamps for.

How to run the client

  1. Make sure API is running.
  2. In Pages > FetchData.razor.cs make sure hubConnection = new HubConnectionBuilder().WithUrl("http://localhost:7071/api").Build(); is set to the port your API is running on.
  3. In the client folder run dotnet run

How it works

The API is powered by Azure Durable Functions. It uses YouTube Explode to get the caption track of the requested YouTube video, slices the caption track into segments, and sends that segment to OpenAI API for a summary. The summary is then broadcasted to SignalR which the client app is connected to.

To do

  • At the moment, it will create 20 segments no matter the length of the video. This is set manually with the slices variable in the API. I plan on making this user setable.
  • Local setup isn't ideal.
    • devcontainer support in the works.
    • azd support in the works.
  • The api folder currently holds a Durable and non durable function. This is because this whole thing orignally started as a console app then I turned it into a HTTPTrigger Azure Function and then Durable Function which is ultimately the best model for this. I will clean this up.
  • Need to improve SignalR connection so API will only broadcast to client and not to all connected.

Contributing

Feel free to open up an issue or reach out to me with.