polly-recorder

A chalice based recording interface for converting text-to-speech using AWS polly.

See: https://sookocheff.com/post/aws/learning-a-language-with-amazon-polly/

Setup

CloudFormation Resources

A CloudFormation template is included to deploy an S3 bucket for storing recordings and a DynamoDB table for indexing the stored recordings.

Create the CloudFormation stack using the AWS CLI setting any parameters correctly.

aws cloudformation create-stack \
    --stack-name polly-recorder \
    --template-body file://cloudformation.yaml \
    --parameters ParameterKey=S3BucketName,ParameterValue=<s3-bucket-name> \
                 ParameterKey=DynamoDBTableName,ParameterValue=<dynamo-table>

Chalice Application

Update .chalice/config.json with environment variables for your application.

• DYNAMO_DB_TABLE: The table used to index recordings.
• S3_BUCKET: The bucket used to store recordings.

Example:

{
  "version": "2.0",
  "app_name": "polly-recorder",
  "stages": {
    "dev": {
      "api_gateway_stage": "dev",
      "environment_variables": {
        "DYNAMO_DB_TABLE": "<your-table-name>",
        "S3_BUCKET": "<your-bucket-name>"
      }
    }
  }
}

Deploy your application:

$ chalice deploy

Update IAM Policies

The Lambda function deployed by Chalice will need to have access to the S3 bucket, the DynamoDB table, and to Amazon Polly. Set the policy on the Lambda function as appropriate.

Create an API Key

Create an API key for API Gateway endpoint to be used by the user interface for (somewhat) secure access.

Web Interface

The web interface (index.html) needs to now which API endpoint and API key to use. Set these in script.js.

Usage

Open index.html to start recording input text.