Skip to main content

Asynchronous wrapper for AWS Polly API

Project description


Python 3.7 PyPi Package Version Amazon Polly API

aiopolly is an asynchronous library for Amazon Polly API which is written with asyncio and aiohttp and uses pydantic models


  • Asynchronous
  • Respects PEP-8 (no camelCase args and vars)
  • Provides easy way to work with SSML tags and lexicons
  • Has mapped and classified AWS API exceptions
  • Has audio convert support and built-in async opus converter


$ pip install aiopolly

Getting started

To work with AWS Polly you need AWS account, IAM User and it's credentials, here's the instructions how to get it

Then you can init this class using one of two methods:

  1. Provide your access and secret keys directly:

    from aiopolly import Polly
    polly = Polly(access_key='your_access_key', secret_key='your_secret_key')
  2. Create a ~/.aws/credentials file with following data:

        aws_access_key_id = your_access_key
        aws_secret_access_key = your_secret_key

    And init class without any auth params:

    from aiopolly import Polly
    polly = Polly()


Many voices

import asyncio
import time

from aiopolly import Polly, types

async def main():
    time_start = time.time()
    # Initializing AWS Polly client with default output_format
    polly = Polly(output_format=types.AudioFormat.mp3)

    voices = await polly.describe_voices()

    text = 'Whatever you can do I can override it, got a million ways to synthesize it'

    # Asynchronously synthesizing text with all available voices
    synthesized = await voices.synthesize_speech(text, language_code=types.LanguageCode.en_US)

    # Asynchronously saving each synthesized audio on disk
    await asyncio.gather(
        *(speech.save_on_disc(directory='examples') for speech in synthesized)

    # Counting how many characters were synthesized
    characters_synthesized = sum(speech.request_characters for speech in synthesized)

    print(f'{characters_synthesized} characters are synthesized on {len(synthesized)}speech'
          f'and saved on disc in {time.time() - time_start} seconds!')

loop = asyncio.get_event_loop()

Managing lexicons

import asyncio

from aiopolly import Polly
from aiopolly.types import Alphabet, AudioFormat, LanguageCode, VoiceID
from aiopolly.utils.lexicon import new_lexicon, new_lexeme

async def main():
    # Creating a new Polly instance with default output format 'mp3'
    polly = Polly(output_format=AudioFormat.mp3)

    text = 'Python is a beautiful programming language which is commonly used for web backend and ML. ' \
           'It also has cool style guides listed in PEP-8, and many community libraries like aiopolly or aiogram.'

    # Creating new lexemes
    python_lexemes = [
        new_lexeme(grapheme='PEP', alias='Python Enhancement Proposals'),
        new_lexeme(grapheme='ML', alias='Machine Learning'),
        new_lexeme(grapheme='aiopolly', phoneme='eɪˈaɪoʊˈpɑli'),
        new_lexeme(grapheme='aiogram', phoneme='eɪˈaɪoʊˌgræm')

    # Creating a new lexicon with 'ipa' alphabet and 'en_US' language code
    lexicon = new_lexicon(alphabet=Alphabet.ipa, lang=LanguageCode.en_US, lexemes=python_lexemes)

    # Putting lexicon on Amazon server
    lexicon_name = 'PythonML'
    await polly.put_lexicon(lexicon_name=lexicon_name, content=lexicon)

    # Synthesizing speech with lexicon we just created 
    # (we don't need to specify required param "output_format", as we using it by default)
    speech = await polly.synthesize_speech(text, voice_id=VoiceID.Matthew, lexicon_names=[lexicon_name])

    # Saving speech on disk with default name
    await speech.save_on_disc()

loop = asyncio.get_event_loop()

Using SSML Text

aiopolly got built-in ssml-text factory which you can use to manage your ssml text:

import asyncio

from aiopolly import Polly
from aiopolly.types import AudioFormat, VoiceID, TextType
from aiopolly.utils.ssml import ssml_text, prosody
from aiopolly.utils.ssml.params import Volume, Pitch, Rate

super_fast = prosody(f'''\
Uh, sama lamaa duma lamaa you assuming I'm a human\
What I gotta do to get it through to you I'm superhuman\
Innovative and I'm made of rubber\
So that anything you say is ricocheting off of me and it'll glue to you\
I'm devastating more than ever demonstrating\
How to give a motherfuckin' audience a feeling like it's levitating\
Never fading and I know that the haters are forever waiting\
For the day that they can say I fell off they'd be  celebrating\
'Cause I know the way to get 'em motivated
    rate=Rate.x_fast, volume=Volume.x_loud, pitch=Pitch.high

async def main():
    # Creating a new Polly instance with default output format 'mp3'
    polly = Polly(output_format=AudioFormat.mp3)

    text = ssml_text(super_fast)

    speech = await polly.synthesize_speech(text, voice_id=VoiceID.Matthew, text_type=TextType.ssml)

    await speech.save_on_disc(directory='speech')

loop = asyncio.get_event_loop()

Using default params

You can init Polly client with any default params. Those will be used when same params in API methods remain empty.

from aiopolly import Polly, types

polly = Polly(
    lexicon_names=['myLexicon', 'alsoMyLexicon'],
    **{'other_default_param': 'value'}

Using built-in OpusConverter

For this to work you need ffmpeg and libopus installed on your system

import asyncio

from aiopolly import Polly
from aiopolly.types import AudioFormat, TextType, VoiceID
from aiopolly.utils.converter import OpusConverter
from aiopolly.utils.ssml import ssml_text, pause, Strength

async def main():
    # Creating instance if OpusConverter
    converter = OpusConverter(auto_convert=True, keep_original=True)

    polly = Polly(output_format=AudioFormat.mp3, converter=converter)

    text = ssml_text(f'''

Use this method to send audio files, if you want Telegram clients to display the file as a playable voice message. 
For this to work, your audio must be in an {pause(Strength.none)}.ogg file encoded with OPUS 
(other formats may be sent as Audio or Document)

    # Synthesizing speech as usual, it will be converted automatically
    speech = await polly.synthesize_speech(text, voice_id=VoiceID.Matthew, text_type=TextType.ssml)

    # Saving speech on disk with default name
    await speech.save_on_disc(directory='speech')
    await speech.save_on_disc(directory='speech', converted=False)

loop = asyncio.get_event_loop()


  • Test Synthesis tasks (not tested yet)
  • Write tests
  • Get rid of botocore (built-in request signer needed)
  • Work on converter API?
  • More docs?
Inspired by Alex Root Junior's aiogram

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

aiopolly-0.2.2.tar.gz (32.8 kB view hashes)

Uploaded source

Built Distribution

aiopolly-0.2.2-py3-none-any.whl (37.1 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page