Audio Transcription App

This is a web application that captures audio using the Web Audio API and streams it to a server via WebSocket for transcription. The user can also send the transcribed text to an AI service for further processing.

Setup

1. Install dependencies:

   npm install

2. Get the AWS tokens and create the .env file:

   AWS_ACCESS_KEY_ID=
   AWS_SECRET_ACCESS_KEY=
   AWS_REGION=us-east-1
   

   Your IAM user needs access to both the Transcribe and Bedrock services. You can create the IAM user using the aws cli:

   aws iam create-user --user-name TranscribeBedrock_User
   aws iam create-policy --policy-name TranscribeBedrock_Policy --policy-document '{
   "Version": "2012-10-17",
   "Statement": [
       {
           "Effect": "Allow",
           "Action": [
               "transcribe:*",
               "bedrock:InvokeModel"
           ],
           "Resource": "*"
       }
   ]
   }'
   aws iam attach-user-policy --user-name TranscribeBedrock_User --policy-arn arn:aws:iam::<your-account-id>:policy/TranscribeBedrock_Policy
   aws iam create-access-key --user-name TranscribeBedrock_User

3. Start the server:

   npm run start