In this example, we’ll build a simple app which transcribes YouTube videos using Whisper, a state-of-the-art model for speech recognition.

​
Setting up the environment

First, you’ll setup your compute environment. You’ll specify:

  • Compute requirements, including a GPU
  • Python and system-level packages to install in the runtime
app.py
app = App(
    name="whisper-example",
    runtime=Runtime(
        cpu=2,
        memory="16Gi",
        gpu="A10G",
        image=Image(
            python_version="python3.8",
            python_packages=[
                "git+https://github.com/openai/whisper.git",
                "pytube @ git+https://github.com/felipeucelli/pytube@03d72641191ced9d92f31f94f38cfb18c76cfb05",
            ],
            commands=["apt-get update && apt-get install -y ffmpeg"],
        ),
    ),
)

​
Transcribing YouTube Videos

We’ll write a basic function which takes in a YouTube video URL, uses the youtube_dl library to download the video as an Output, and runs the video through Whisper to generate a text transcript.

​
Deployment

You’ll deploy the app by entering your shell, and running:

beam deploy app.py

Your Beam Dashboard will open in a browser window, and you can monitor the deployment status in the web GUI.

​
Calling the API

You’ll call the API by pasting in the cURL command displayed in the browser window.

curl -X POST --compressed "https://apps.beam.cloud/77n0o" \
   -H 'Authorization: [YOUR_AUTH_TOKEN]' \
   -H 'Content-Type: application/json' \
   -d '{"video_url": "https://www.youtube.com/watch?v=adJFT6_j9Uk"}'

The API will return a transcript with our video:

{
  "pred": " Welcome to the Pets Show. That is, Physics Explained in 10 Seconds. For the next month, in addition to Minute Physics, I'll be making one 10-second video every day. 10 Seconds of Physics explaining 5 seconds of titles on either end."
}