Radio reporter

Radio reporter listens to your favourite streaming radio stations and identifies and reports on what songs are being played.

Songs have quite a bit of metadata associated with them so I wrote this to build a dataset that I could visualise. I'll also be able to build dope playlists with it.

Requirements

This application is designed to live on AWS (requires account) and requires Python 3, Node.js, and Serverless

• Python
• Node.js
• Serverless Framework

You will need to register as a spotify developer and create an app here

You will also need to deploy ffmpeg-lambda-layer to your AWS account. You can do that here

Environment variables

You will need to create a copy of .env.sample in the root directory and rename it to .env before updating the values as you want them

• STREAM_URLS - comma separated list of radio stream urls
• SPOTIPY_CLIENT_ID - the client id from your spotify app
• SPOTIPY_CLIENT_SECRET - the client secret from your spotify app
• FFMPEG_ARN - the arn from your ffmpeg-lambda-layer deployment

Build the dependency layers

As the layers are already included in this repo, this only needs to be done if you want to update the package versions

docker build -f Dockerfile -t reporterlayers:latest .
docker run -dit reporterlayers:latest
docker cp {container_id}:/var/task/recogniselayer.zip .
docker cp {container_id}:/var/task/dotenvlayer.zip .

Deploy to AWS

sls deploy

Remove the service

sls remove

Future enhancements

• Remove the dependency on s3 to reduce cost (send data direct to lambda)
• Move the stream url list from env to json file to allow extra metadata (station name etc)
• Add extra access patterns to DynamoDB (artist, show)
• Add visualisation dashboard