sakaru/podcastify

Make a podcast from some textual source using fancy lambdas

Podcastify

Make a podcast out of any text; an RSS feed, a news website, product reviews

Flow

1. Triggered by a scheduler, Ingestor lamda reads the input.
2. Schedules Text-To-Speach tasks.
3. Save into DynamoDB: - PRI=task_id | status="waiting_for_polly"
4. Text-To-Speach engine runs. Outputs .ogg files into an S3 bucket.

• Also triggers SNS

3. SNS triggers Polly Lambda
4. If a Polly job failed, schedule a retry. Delete the DynamoDB record. Create a new one.
5. If the Polly job succeeded, update status to "waiting_for_merge"
6. Triggered by Scheduler, Merger Lambda starts.
7. Scans DynamoDB for jobs with status="waiting_for_polly". If found, Exit
8. Scans DynamoDB for jobs with status="waiting_for_merge"
9. (re-)create RSS (is idempotent)