Skip to main content

AWS Polly

Amazon Polly is a cloud-based text-to-speech service that offers high-quality voice synthesis with support for multiple languages and voices.

Authentication​

To use AWS Polly, you'll need AWS credentials:

from tts_wrapper import PollyTTS, PollyClient

client = PollyClient(credentials=(
    'aws_region',      # e.g., 'us-east-1'
    'aws_key_id',      # Your AWS Access Key ID
    'aws_secret_key'   # Your AWS Secret Access Key
))

tts = PollyTTS(client)
tip

Use environment variables or AWS credentials file for secure credential management:

import os

client = PollyClient(credentials=(
    os.getenv('AWS_REGION'),
    os.getenv('AWS_ACCESS_KEY_ID'),
    os.getenv('AWS_SECRET_ACCESS_KEY')
))

Features​

SSML Support​

AWS Polly provides comprehensive SSML support:

ssml_text = """
<speak>
    Hello <break time="300ms"/> World!
    <prosody rate="slow" pitch="+20%">
        This is a test of SSML features.
    </prosody>
</speak>
"""
tts.speak(ssml_text)

See Amazon Polly SSML Reference for all supported tags.

Streaming​

Polly supports real-time audio streaming:

# Stream synthesis for real-time playback
tts.speak_streamed("This text will be synthesized and played in real-time")

Word Timing​

Get precise timing information for each word:

def on_word(word: str):
    print(f"Speaking: {word}")

tts.connect("started-word", on_word)
tts.speak("This text will trigger word timing callbacks")

Voice Selection​

List and select from available voices:

# Get list of available voices
voices = tts.get_voices()
for voice in voices:
    print(f"ID: {voice['Id']}, Language: {voice['LanguageCode']}")

# Set a specific voice
tts.set_voice("Joanna", "en-US")

Best Practices​

  1. Cost Management

    • Use streaming for long text to optimize bandwidth
    • Cache frequently used phrases
    • Monitor usage through AWS Console

  2. Performance

    • Reuse client instances
    • Use appropriate sampling rates
    • Consider regional endpoints for lower latency

  3. Error Handling

    try:
        tts.speak("Hello, world!")
    except Exception as e:
        if "CredentialsError" in str(e):
            print("Check your AWS credentials")
        elif "QuotaExceeded" in str(e):
            print("AWS Polly quota exceeded")
        else:
            print(f"Error: {e}")

Limitations​

  • Maximum text length of 3000 characters per request
  • API rate limits apply (check AWS quotas)
  • Certain SSML features limited to specific voices
  • Neural voices not available in all regions

Additional Resources​

Next Steps​