Amazon Polly

AWS text to speech billed per character

by Amazon Web Services

Overview

Amazon Polly is AWS's text-to-speech service and it is billed by the character rather than by subscription: you reach it from the AWS console or from your own code, and AWS states that "You are billed monthly for the number of characters of text that you processed", with Standard voices at $4.00 per 1 million characters and Neural voices at $16.00 (aws.amazon.com/polly/pricing, read 12 October 2026). The console is a real entry point - AWS describes logging in and typing text to "listen to generated speech or save it as an audio file" - but there is no editor, app or studio here: the same site states that Polly is a fully-managed cloud service that you communicate with through APIs in your code and that you cannot download or deploy its source code. The pricing page publishes the conversion a narration budget needs: a worked example lists 1 million characters as about 23 hours and 8 minutes of speech at the same $4.00, and another row prices a typical news article of about 6,500 characters at about 9 minutes for $0.03. The hard limit is per request rather than per month: the SynthesizeSpeech API accepts at most 6,000 characters in total, of which no more than 3,000 are billed characters, and the asynchronous StartSpeechSynthesisTask raises those ceilings to 200,000 and 100,000.

No single starting price is published here, because the page does not support one. It prints amounts with a dollar sign and no currency code, and it prices the same voices two ways: the headline table gives $4.00 and $16.00 per million characters, while the AWS GovCloud (US) section gives $4.80 and $19.20 for the same Standard and Neural voices, without naming the region the headline table applies to - so the amount depends on where you run it, and this entry records the rate per million characters instead of a plan price. The pricing page lists service free allowances, including 5 million Standard characters monthly, alongside separate rules for new AWS customers from 15 July 2025: up to $200 credits, a six-month Free plan, and credits expiring within twelve months of account creation. These are conditional offers, not simultaneous or perpetual benefits for every account. New AWS customers receive USD100 at signup and may earn up to USD100 more through activities. The Free account plan ends after six months or credit depletion, whichever comes first; paid accounts incur charges beyond eligible credits. Account eligibility, remaining credit and expiry must be checked before treating a Polly job as free. What the console itself exports is not established either, since the answer about saving an audio file names no format; the documented containers come from the API reference, which lists mp3, ogg_vorbis, ogg_opus, mu-law, a-law or pcm - WAV is not among them.

The recorded ownership and reuse statements do not settle every commercial-use condition. AWS states: "As between you and AWS, your Polly output belongs to you", and the service is described as having no restrictions on storing and reusing generated speech, with static prompts replayed repeatedly and mass-notification use explicitly not restricted either. Training and data use is answered as an opt-out: AWS states that text inputs may be stored and used to improve and develop the quality of Polly and other AWS machine-learning technologies, "including the development and training of related technologies", and that an AWS Organizations opt-out policy is how you stop it; content is stored encrypted in the region where you use the service, and opting out also stops copies being stored in another region. Whether an API is included in a consumer plan has no purchase to point at, because there is no consumer plan: Polly is the API, billed per character against an AWS account, and the SDKs, HTTP API and CLI are the way in.

Editor's take

Choose it when the deciding number is the cost of a finished minute rather than a monthly plan, and when driving the service from code or the AWS console is acceptable. The conversion this kind of brief needs is published on the pricing page instead of being left to the reader: a million characters is about 23 hours of speech for $4.00 on Standard voices, which puts half an hour of narration in the region of a few cents, but free coverage depends on the account’s offer, balance and expiry. Budget the work rather than the money, because the per-request ceiling means a thirty-minute piece is several requests or one asynchronous job, and read the rate as region-dependent, since the same page prices the same voices higher in GovCloud. There is also no studio: assembling the clips is your side of the job.

Our reading of who this suits, based on the sources above. Opinion, offered to help you decide - not a lab result.

Key Features

•Standard, Neural, Long-Form and Generative voice engines at published per-million-character rates
•100+ voices in 40+ languages, with SSML, custom lexicons and speech marks
•Output containers listed as mp3, ogg_vorbis, ogg_opus, mu-law, a-law or pcm, at sample rates up to 48kHz
•Conditional free offers; check account eligibility, balance and expiry before budgeting a free job
•AWS console entry point that can save generated speech as an audio file, alongside SDKs, CLI and HTTP API
•Content stored encrypted in the region you use, with an AWS Organizations opt-out for the improvement use

Use Cases

→Narrating a course module or an explainer video from a script
→Generating fixed prompts and announcements that are replayed many times
→Adding spoken output to an application through the AWS SDKs
→Producing audio at a published cost per minute instead of buying a monthly plan

Pros

✓The vendor publishes the characters-to-duration conversion, so a narration can be costed before it is generated
✓Your Polly output belongs to you, and AWS states there are no restrictions on storing and reusing generated speech
✓Billing is per character with no subscription, so a small job does not buy a month of unused quota
✓Containers, sample rates and per-request character limits are all documented in the API reference

Cons

✕No single starting price is published here, because the same voices cost more in GovCloud and the page does not name the region its headline table applies to
✕There is no editor, app or studio: the entry points are the AWS console and your own code
✕A thirty-minute narration exceeds the 3,000 billed characters one SynthesizeSpeech request accepts, so it takes several requests or an asynchronous task
✕Text inputs may be used to improve and train AWS machine-learning technologies unless the account opts out, and WAV is not among the documented containers

Frequently Asked Questions

How much does Amazon Polly cost?

Polly bills monthly by processed characters. Its headline per-million-character rates are $4 Standard, $16 Neural, $100 Long-Form and $30 Generative; GovCloud Standard/Neural rates differ. The table does not establish this reader’s region-specific final bill. New AWS customers receive USD100 at signup and may earn up to USD100 more through activities. The Free account plan ends after six months or credit depletion, whichever comes first; paid accounts incur charges beyond eligible credits. Account eligibility, remaining credit and expiry must be checked before treating a Polly job as free. Free usage depends on account eligibility, offer type, remaining balance and expiry; it is not guaranteed to cover this thirty-minute task.

How many minutes of audio is a million characters?

The vendor's own pricing example states that 1 million characters is about 23 hours and 8 minutes of speech, and lists a typical news article of about 6,500 characters at about 9 minutes. By that conversion a thirty-minute narration sits in the region of 21,600 characters, which is our arithmetic on the vendor's figure rather than a statement from AWS.

May I reuse the generated speech commercially?

AWS records ownership of your Polly output and permits storage and reuse, while requiring rights to third-party input. Those statements do not settle every commercial-use condition for your project; this catalogue keeps that broader permission unknown.

Is my written text used to train AWS models?

It may be, unless the account opts out. AWS states that text inputs may be stored and used to provide and maintain the service and to improve and develop the quality of Polly and other AWS machine-learning technologies, including the development and training of related technologies, and that an AWS Organizations opt-out policy stops that use; content is encrypted and stored at rest in the region where you use Polly, and opting out also prevents storage in another region.

Which formats and length limits apply?

The API reference lists the output container as mp3, ogg_vorbis, ogg_opus, mu-law, a-law or pcm, with sample rates up to 48kHz for mp3 and ogg_vorbis, so WAV is not offered. One SynthesizeSpeech request accepts at most 6,000 characters of which no more than 3,000 are billed, and the asynchronous StartSpeechSynthesisTask accepts up to 200,000 with 100,000 billed. The FAQ answer about saving an audio file from the console names no format, so the console export format is not established here.

Task guides for this product

T12 · Choose a text-to-speech route for deliverable narration

Key facts and where they come from

PriceNot established: Published per-character rates and conditional free offers do not establish the selected account’s region, engine, active benefit, remaining credits, expiry or final bill.Source: Amazon Polly — pricing and conditional free offers · AWS — Free and Paid account plans
AvailabilityNot established: Country-by-country distribution and actual reachability of the selected product and plan are not established by the registered sources. A platform list, hosting location or payment-region note does not establish access from a particular country.Source: Amazon Polly - FAQs
Commercial useNot established: The FAQ establishes output ownership and requires third-party input rights, but does not explicitly establish the full commercial-use licence; ownership and a media use case do not settle every commercial condition.Source: Amazon Polly - FAQs · Amazon Polly - product page
Training on your dataDocumented: 'Amazon Polly may store and use text inputs processed by the service solely to provide and maintain the service and to improve and develop the quality of Amazon Polly and other Amazon machine-learning/artificial-intelligence technologies … including the development and training of related technologies.' / 'You may opt out of having your content used to improve and develop the quality of Amazon Polly … by using an AWS Organizations opt-out policy.Source: Amazon Polly - FAQs
API billingDocumented: Polly bills monthly by processed characters. Its headline per-million-character rates are $4 Standard, $16 Neural, $100 Long-Form and $30 Generative; GovCloud Standard/Neural rates differ. The table does not establish this reader’s region-specific final bill. The pricing page lists service free allowances, including 5 million Standard characters monthly, alongside separate rules for new AWS customers from 15 July 2025: up to $200 credits, a six-month Free plan, and credits expiring within twelve months of account creation. These are conditional offers, not simultaneous or perpetual benefits for every account. Its illustration equates one million characters with about 23 hours and eight minutes; this is an estimate, not a completed-file test. New AWS customers receive USD100 at signup and may earn up to USD100 more through activities. The Free account plan ends after six months or credit depletion, whichever comes first; paid accounts incur charges beyond eligible credits. Account eligibility, remaining credit and expiry must be checked before treating a Polly job as free.Source: Amazon Polly — pricing and conditional free offers · AWS — Free and Paid account plans
Export formatsDocumented: 'OutputFormat — The format in which the returned output will be encoded. For audio stream, this will be mp3, ogg_vorbis, ogg_opus, mu-law, a-law or pcm. For speech marks, this will be json.' Valid values are listed as 'json | mp3 | ogg_opus | ogg_vorbis | pcm | mulaw | alaw'. Sample rates for mp3 and ogg_vorbis are '8000', '16000', '22050', '24000', '44100' and '48000'. TextLengthExceededException: 'For the SynthesizeSpeech API, the limit for input text is a maximum of 6000 characters total, of which no more than 3000 can be billed characters. For the StartSpeechSynthesisTask API, the maximum is 200,000 characters, of which no more than 100,000 can be billed characters.' WAV is not among the formats.Source: Amazon Polly - SynthesizeSpeech API reference
Usage limitsDocumented: 'OutputFormat — The format in which the returned output will be encoded. For audio stream, this will be mp3, ogg_vorbis, ogg_opus, mu-law, a-law or pcm. For speech marks, this will be json.' Valid values are listed as 'json | mp3 | ogg_opus | ogg_vorbis | pcm | mulaw | alaw'. Sample rates for mp3 and ogg_vorbis are '8000', '16000', '22050', '24000', '44100' and '48000'. TextLengthExceededException: 'For the SynthesizeSpeech API, the limit for input text is a maximum of 6000 characters total, of which no more than 3000 can be billed characters. For the StartSpeechSynthesisTask API, the maximum is 200,000 characters, of which no more than 100,000 can be billed characters.' WAV is not among the formats.Source: Amazon Polly - SynthesizeSpeech API reference

Official sources

The vendor pages we read for the claims on this page, what each one states, and the day we read it.

Amazon Polly — pricing and conditional free offers ↗
Polly bills monthly by processed characters. Its headline per-million-character rates are $4 Standard, $16 Neural, $100 Long-Form and $30 Generative; GovCloud Standard/Neural rates differ. The table does not establish this reader’s region-specific final bill. The pricing page lists service free allowances, including 5 million Standard characters monthly, alongside separate rules for new AWS customers from 15 July 2025: up to $200 credits, a six-month Free plan, and credits expiring within twelve months of account creation. These are conditional offers, not simultaneous or perpetual benefits for every account. Its illustration equates one million characters with about 23 hours and eight minutes; this is an estimate, not a completed-file test.
Read on October 15, 2026
Amazon Polly - FAQs ↗
'Simply login to your AWS account and navigate to the Amazon Polly console (which is a part of the AWS Console). You can then use the console to type in any text and listen to generated speech or save it as an audio file.' / 'Amazon Polly supports MP3, Vorbis, and raw PCM audio stream formats.' / 'Who owns Polly recordings? As between you and AWS, your Polly output belongs to you. If you input text into Polly that belongs to a third party, we require that you have the rights to do so.' / 'Do I still own my content that is processed and stored by Amazon Polly? You always retain ownership of your content and we will only use your content with your consent.' / 'Amazon Polly may store and use text inputs processed by the service solely to provide and maintain the service and to improve and develop the quality of Amazon Polly and other Amazon machine-learning/artificial-intelligence technologies … including the development and training of related technologies.' / 'You may opt out of having your content used to improve and develop the quality of Amazon Polly … by using an AWS Organizations opt-out policy.' / 'Any content processed by Amazon Polly is encrypted and stored at rest in the AWS region where you are using Amazon Polly.' / Regions answered with the AWS Regional Services List plus a named subset for Neural voices. / 'Can I use the service for generating static voice prompts that will be replayed multiple times? Yes, you can. The service does not restrict this'. / 'Yes, as part of the AWS Free Usage Tier … new Amazon Polly customers can synthesize millions of characters for free each month for the first 12 months.'
Read on October 12, 2026
Amazon Polly - SynthesizeSpeech API reference ↗
'OutputFormat — The format in which the returned output will be encoded. For audio stream, this will be mp3, ogg_vorbis, ogg_opus, mu-law, a-law or pcm. For speech marks, this will be json.' Valid values are listed as 'json | mp3 | ogg_opus | ogg_vorbis | pcm | mulaw | alaw'. Sample rates for mp3 and ogg_vorbis are '8000', '16000', '22050', '24000', '44100' and '48000'. TextLengthExceededException: 'For the SynthesizeSpeech API, the limit for input text is a maximum of 6000 characters total, of which no more than 3000 can be billed characters. For the StartSpeechSynthesisTask API, the maximum is 200,000 characters, of which no more than 100,000 can be billed characters.' WAV is not among the formats.
Read on October 12, 2026
Amazon Polly - product page ↗
'Amazon Polly is a fully-managed service that generates voice on demand, converting any text to an audio stream.' / 'Integrate the Amazon Polly API into your existing applications to become voice-ready quickly.' / 'Create voiceovers for animations, games, and other media directly from your scripts.' / 'Amazon Polly offers 100+ male and female voices in 40+ language and language variants.' / 'No. Amazon Polly is a fully-managed cloud AI service. You communicate with it using APIs in your code. You cannot download or deploy Amazon Polly source code in your environment.'
Read on October 12, 2026
AWS — Free and Paid account plans ↗
New AWS customers receive USD100 at signup and may earn up to USD100 more through activities. The Free account plan ends after six months or credit depletion, whichever comes first; paid accounts incur charges beyond eligible credits. Account eligibility, remaining credit and expiry must be checked before treating a Polly job as free.
Read on October 15, 2026

Compare with other tools in Audio Generation

Product	Maker	Pricing	Launched	Shared tags
ElevenLabs	ElevenLabs	Freemium	2023-01	Voice, Text-to-speech
Typecast	Neosapience	Freemium	Launch date not verified	Voice, Text-to-speech
Cartesia	Cartesia AI, Inc.	Freemium	Launch date not verified	Voice, Text-to-speech
Udio	Udio	Paid	2024-04-10	—
Suno	Suno	Freemium	2023-12-20	—
Stable Audio	Stability AI	Freemium	2023	—