Skip to main content

This is a prototype vision of how a future government service could work. It's not a real service yet, but we're exploring what it could look like. Your feedback will help shape the real service.

Step 1: Access the Polly Interface - Text-to-Speech Walkthrough

Open the Amazon Polly text-to-speech console

Walkthrough progress

Step 1 of 4 • 2 minutes

Step 1 2 minutes

Access the Polly Interface

Find your deployed Text-to-Speech resources and open the Amazon Polly console.

Find your Text to Speech application URL in CloudFormation Outputs
The application is ready to convert text into natural speech

Expected outcome

  • CloudFormation stack 'ndx-try-text-to-speech-[timestamp]' shows CREATE_COMPLETE status
  • Polly console URL opens successfully
  • Text input area and voice selection visible
  • You're ready to convert text to speech

Find your deployment

The Text-to-Speech scenario deployed via CloudFormation. Let's find your stack and access the Polly console:

  1. Open the CloudFormation console

    Go to CloudFormation console (us-east-1 N. Virginia)

  2. Find your stack

    Look for a stack named ndx-try-text-to-speech-[timestamp] with status CREATE_COMPLETE

  3. Click the Outputs tab

    Select your stack, then click the "Outputs" tab. You'll see a list of resources created by the deployment.

  4. Open the Polly console URL

    Find the output key PollyConsoleURL and click the URL value. This opens the Amazon Polly text-to-speech interface with pre-configured permissions.

What's happening: The CloudFormation stack created an S3 bucket for audio storage and configured IAM permissions for Polly access. The console URL includes embedded credentials so you don't need to manage AWS permissions manually.

What you should see

The Amazon Polly console interface includes:

Text input area
Large text box where you paste or type council announcements (up to 3000 characters)
Engine selection
Dropdown to choose "Neural" (recommended) or "Standard" engine
Language selection
Dropdown showing "British English (en-GB)" pre-selected
Voice selection
Dropdown listing Amy, Brian, Emma (UK English neural voices)
Listen button
Orange "Listen" button to generate and play audio preview
Download to S3
Blue "Synthesize to S3" button to save audio file permanently

Verify your access

Quick test to ensure Polly is working:

  1. In the text input area, type: Testing Amazon Polly text to speech
  2. Verify "Engine" is set to Neural (not Standard)
  3. Verify "Language" is set to British English
  4. Select voice Amy from the dropdown
  5. Click the orange "Listen" button
  6. You should hear Amy's voice say the test sentence in ~2 seconds

Did you hear the audio? If yes, Polly is working correctly. If not, check the troubleshooting section below.

Understanding the interface

Neural vs Standard engine

Neural: Uses deep learning to generate natural-sounding speech with emotion and inflection. Sounds like a real person. Costs £0.016 per 1000 characters.

Standard: Older text-to-speech technology. Sounds robotic. Costs £0.004 per 1000 characters (4× cheaper but noticeably lower quality).

Recommendation: Always use Neural for public-facing council communications. The quality difference is worth the small additional cost (£0.008 per typical announcement).

Listen vs Synthesize to S3

"Listen" button: Generates audio on-demand for preview. Audio is not saved - use this for testing different voices or making text edits.

"Synthesize to S3" button: Generates audio and saves MP3 file to your S3 bucket with a permanent URL. Use this when you're ready to publish the audio on your website or phone system.

Recommendation: Use "Listen" to compare voices and finalize text, then "Synthesize to S3" once to save the final version. Avoid repeatedly saving to S3 during testing.

Troubleshooting

CloudFormation stack not found or status is not CREATE_COMPLETE

If you don't see the Text-to-Speech stack:

  • Check you're in the correct AWS region (us-east-1 N. Virginia)
  • Verify you deployed the scenario (go to Text-to-Speech scenario page)
  • If stack shows CREATE_IN_PROGRESS, wait 2-3 more minutes for completion
  • If stack shows CREATE_FAILED, delete and redeploy (check CloudWatch logs for error details)
Polly console URL shows "Access Denied" error

If you can't access the Polly console:

  • Verify you're logged into the same AWS account where you deployed the stack
  • Check the URL is complete (should be very long with embedded credentials)
  • Try copying the URL again from CloudFormation Outputs (may have been truncated)
  • Ensure IAM permissions were created correctly (check CloudFormation Events tab for errors)

If still blocked, go directly to the Polly console (direct link) and manually configure.

"Listen" button does nothing or shows error

If audio doesn't play when you click Listen:

  • Check your browser allows audio playback (may need to enable in browser settings)
  • Verify text input is not empty (Polly can't generate audio from blank text)
  • Try selecting a different voice (some voices may fail if misconfigured)
  • Check browser console (F12) for JavaScript errors
  • Try a different browser (Chrome, Firefox, Safari all support Polly)
Only seeing "Standard" voices, not "Neural" voices

If Amy, Brian, Emma don't appear in the voice dropdown:

  • Ensure "Engine" dropdown is set to Neural (not Standard)
  • Ensure "Language" is set to British English (en-GB) not US English
  • Neural voices may not be available in all AWS regions - confirm you're in us-east-1 (N. Virginia)
  • If using Polly API directly, verify you have neural engine permissions in IAM

Once you can hear Amy's voice in the test, you're ready to convert real council announcements.

Need help?

If you're stuck, check the troubleshooting section or contact us.