Low Code AI Tutorial - Automated Transcription (Speech-to-Text) with Bubble.io and AWS Transcribe

Learn to build an automated transcription platform, using Bubble.io and Amazon Web Services, while only writing about 30 lines of code!

We build an application that allows users to upload audio files and generate transcriptions for them with the click of a button.

We use the following tools:
Bubble.io for the front-end
AWS Transcribe - Amazon's Speech-to-Text API
AWS Serverless tools (Lambda and API Gateway) to handle API processing.

There is a little bit more code in this tutorial than in previous ones, but its still fairly minimal, considering the functionality we end up with.

-- Timestamps --
0:00 - 4:40 Intro & Overview
4:40 - 8:00 AWS Transcribe Overview
8:00 - 9:50 Alternatives to AWS Transcribe
9:50 - 16:50 Build front-end in Bubble (1)
16:50 - 19:15 Creating shortened audio files
19:15 - 30:10 Build front-end in Bubble (2)
30:10 - 1:03:55 Build back-end in AWS
1:03:55 - 1:31:56 Connect Bubble & AWS, and Automate Updates

You can find the code used in the tutorial, for the AWS Lambda functions, here: https://github.com/Launchable-AI/low_code_transcription_tutorial

Note - there are a few Bubble inaccuracies in here. e.g., at ~14:30 you don't really need to create the "Audios" field for the User, and track uploads that way; the built-in "Created By" property can work for the same purpose, and around 28:00, you could use auto-binding instead of a workflow. Also, the method of repeatedly calling the GetTranscriptionJob API endpoint is not ideal; you could set up a trigger and a webhook (via a Bubble backend workflow that is public), and use that to cause your Bubble app to go out and grab the file). This is slightly more involved, but a cleaner pattern, if you can hack it :)

Feel free to leave any other improvements/optimizations you see in the comments :)

IMPORTANT: If you are using the default Bubble database for your S3 files (as demonstrated in the video), you must deploy your API and Lambda functions to us-east-1, to match the region of Bubbe S3 bucket. I forgot to mention this in the video, and a couple of viewers have gotten tripped up.

If you like this content, let us know! Leave a comment if there is a specific API or machine learning application you would be interested in seeing.

Thanks for watching!

Easter egg hunt - how many times do I say "in a nutshell" in a this video?
Also, please pardon some editing oversights :) e.g., repeating the explanation of the ffmpeg command

-------------------------------------------
website: launchable.ai
twitter - @LaunchableAI - https://twitter.com/LaunchableAI
LaunchableAI - Helping Makers, Founders, and Businesses launch AI-powered products and businesses. We offer development services for AI/ML, no-code/low-code, data science, and web applications.

Видео Low Code AI Tutorial - Automated Transcription (Speech-to-Text) with Bubble.io and AWS Transcribe канала Launchable AI