HI, This workflow reads my subscribed technical blogs in my gmail inbox, generates a short summary (around 200 words), converts that to an audio file and sends me back. This helps me focus on blogs which I really want to read and avoid the promotional ones and saves me time.
I have used Open AI for both summary and audio file generation and it is taking around 10-15 second to generate a 1 min long audio file. This is just a barebone structure which is working fine for happy path.
Please suggest me improvements to make it a full proof solution which can be published as a template for use.
GmailTrigger: Gets triggered when an email is received from subscribed blogs. I have labels for each blog and I have used Gmail ‘search’ based on labels.
Filter: This node is used to narrow down the output from previous node as per my requirement.
HTML1: Used to extract the text body from HTM email content
OpenAI: Used to generate summary of 200 words. model-GPT-3.5Turbo
OpenAI1: This node generates the audio file from summary.model-TTS-1
Gmail1:Used to send the audio file as attachment to my gmail account
Gmail:Used to mark the emails as read where audio has been generated and email sent.