Amazon Transcribe

Hands-On

Demo

In this demo, we will perform the following tasks sequentially:

Create S3 Buckets for Input Audio and Output Transcripts.
Upload the Sample Audio File to the Input S3 Bucket.
Create and Run an Amazon Transcribe Job.
Test the Setup: Review and Verify the Transcription Output.
Clean Up AWS Resources to Avoid Charges.

Agenda

Visual Representation

Create Input bucket

transcribe-input-137139

Create Output bucket

transcribe-output-137139

https://archive.org/download/Greatest_Speeches_of_the_20th_Century/FarewelltoBaseball-1939.mp3

Download the Sample Audio File

Upload

Upload succeeded

Amazon Transcribe

Transcription jobs

my-first-transcription-job

Model type

Input data

Output data

Tags - optional

Configure job

Content removal

Transcription job successfully created

Complete

Application integration

transcribe-output

my-first-transcription-job.json

Real-time Transcription

Amazon Transcribe can process speech at a rate of sixteen thousand samples per second with less than three second latency.

This technology is perfect for customer service calls, live meetings, and voice-controlled applications in 
enterprise environments.

Notice how it handles punctuation, capitalization, and even technical terms like EC2, S3, and DynamoDB with high accuracy.

Now I'll pause... [have someone else speak] This demonstrates how Transcribe can identify different speakers in 
real-time conversations.

Let's test some challenging words: Kubernetes, API Gateway, machine learning, artificial intelligence, 
and AWS Lambda functions.

This real-time transcription capability makes Amazon Transcribe ideal for live applications requiring immediate 
speech-to-text conversion.