# Amazon Polly voice engines
Amazon Polly has four voice engines that convert input text into life-like speech. These include: **Generative**, **Long-form**, **Neural**, and **Standard**. To use an Amazon Polly voice, select an engine and a speech synthesis API operation. Then provide input text for the engine to synthesize, and select an audio output format. Given these inputs, Amazon Polly synthesizes the provided text into a high-quality speech audio stream.
The following sections include details about the voice engines offered by Amazon Polly.
**Topics**
+ [Generative voices](generative-voices.md)
+ [Long-form voices](long-form-voices.md)
+ [Neural voices](neural-voices.md)
+ [Standard voices](standard-voices.md)
+ [Choosing a voice engine](using-voices.md)
# Generative voices
Amazon Polly's **generative** text-to-speech (TTS) engine offers the most human-like, emotionally engaged, and adaptive conversational voices available for the use via the Amazon Polly console.
The **Generative engine** is the largest Amazon Polly TTS model to-date. It deploys a billion-parameter transformer that converts raw text into speech codes, followed by a convolution-based decoder that converts these speech codes into waveforms in an incremental, streamable manner. This method shows the widely-reported emergent abilities of Large Language Models (LLMs) when trained on increasing volumes of publicly available and proprietary data comprising a variety of voices, languages, and styles.
The Generative engine creates synthetic speech which is emotionally engaged, assertive, and highly colloquial in a way that is remarkably similar to a human voice. You can use these voices as a knowledgeable customer assistant, a virtual trainer, or an advertiser with a near-human synthetic speech.
**Note**
The state-of-the-art technology underlying these voices falls within the paradigm of generative AI for language and voice modelling. A side effect of the technology is that any updates to the training data and the model could result in slight variations to the way the voices sound, even in case when their overall quality improves with model updates. This could have an impact on use cases with different content parts synthesized over a long time period – for example, a season of podcasts.
## Available generative voices
Amazon Polly currently offers 43 voices in a generative variant.
| | Language | Language code | Name/ID | Gender |
| --- | --- | --- | --- | --- |
| 1 | **English (Australian)** | en-AU | Olivia | Female |
| 2 | **English (British)** | en-GB | Amy Brian | Female Male |
| 3 | **English (Indian)** | en-IN | Kajal | Female |
| 4 | **English (Ireland)** | en-IE | Niamh | Female |
| 5 | **English (New Zealand)** | en-NZ | Aria | Female |
| 6 | **English (Singaporean)** | en-SG | Jasmine | Female |
| 7 | **English (South African)** | en-ZA | Ayanda | Female |
| 8 | **English (US)** | en-US | Danielle Joanna Matthew Ruth Salli Stephen Tiffany | Female Female Male Female Female Male Female |
| 9 | **Dutch (Belgium)** | nl-BE | Lisa | Female |
| 10 | **Dutch (Netherlands)** | nl-NL | Laura | Female |
| 11 | **French (Belgian)** | fr-BE | Isabelle | Female |
| 12 | **French (Canadian)** | fr-CA | Gabrielle Liam | Female Male |
| 13 | **French (France)** | fr-FR | Ambre Céline Florian Léa Rémi | Female Female Male Female Male |
| 14 | **German (Austria)** | de-AT | Hannah | Female |
| 15 | **German (Germany)** | de-DE | Daniel Lennart Vicki | Male Male Female |
| 16 | **German (Swiss)** | de-CH | Sabrina | Female |
| 17 | **Italian (Italy)** | it-IT | Beatrice Bianca Lorenzo | Female Female Male |
| 18 | **Korean (Korea)** | ko-KR | Seoyeon | Female |
| 19 | **Polish (Poland)** | pl-PL | Ewa Ola | Female Female |
| 20 | **Portuguese (Brazilian)** | pt-BR | Camila | Female |
| 21 | **Spanish (Mexican)** | es-MX | Andrés Mía | Male Female |
| 22 | **Spanish (Spain)** | es-ES | Lucia Sergio | Female Male |
| 23 | **Spanish (US)** | es-US | Lupe Pedro | Female Male |
**Note**
Generative voices cost is specified on the [Amazon Polly pricing information page](https://aws.amazon.com/polly/pricing/).
## Feature and region compatibility
Amazon Polly generative voices are available in the following regions:
+ US East (N. Virginia): us-east-1
+ Europe (Frankfurt): eu-central-1
+ US West (Oregon): us-west-2
+ Asia Pacific (Tokyo): ap-northeast-1
+ Asia Pacific (Seoul): ap-northeast-2
+ Asia Pacific (Singapore): ap-southeast-1
+ Europe (London): eu-west-2
+ Canada (Central): ca-central-1
+ Other Regions are not available
**The following features are supported for generative voices:**
+ Bidirectional Streaming API is now offered in Generative engine and allows for streaming input and output at the same time. This API is available in the following AWS regions: US East (N. Virginia), Europe (Frankfurt), US West (Oregon), and Asia Pacific (Singapore). Visit the [documentation](https://docs.aws.amazon.com/polly/latest/dg/API_StartSpeechSynthesisStream.html) to learn more about how to use it.
+ Real-time and asynchronous speech synthesis operations.
+ Newscaster speaking style is not supported in the **Generative** engine.
+ Many (but not all) SSML tags are supported by Amazon Polly. For more information about NTTS-supported SSML tags, see [Supported SSML tags](https://docs.aws.amazon.com/polly/latest/dg/supportedtags.html)
+ As with standard voices, you can choose from various sampling rates to optimize the bandwidth and audio quality for your application. Valid sampling rates for standard and neural voices are 8 kHz, 16 kHz, 22 kHz, or 24 kHz. The default for standard voices is 22 kHz. The default for generative voices is 24 kHz. Amazon Polly supports MP3, OGG (Vorbis), and raw PCM audio stream formats.
*Support for generating speech marks is currently not available.*
**Note**
Currently Europe (London) and Canada (Central) regions only support the following Generative voices: Tiffany (en-US), Amy (en-GB), Brian (en-GB), Florian (fr-FR), Ambre (fr-FR), Lorenzo (it-IT), Beatrice (it-IT), Jasmine (en-SG), Aria (en-NZ), Sabrina (de-CH), Hannah (de-AT), Niamh (en-IE), Camila (pt-BR), Lisa (nl-BE), and Seoyeon (ko-KR)
**Note**
In the unlikely event of model hallucination, (and with the Generative engine's model behavior of rendering the speech token by token) an imposed emergency stop mechanism is in place. The built-in mechanism stops the model from rendering speech any further. This safety feature is based on data analysis where the model has the potential to hallucinate, usually at the end of the sentence.
There could be cases where the model thinks it is going to hallucinate and then might end up cutting a word during a generation step, thus rendering half the word. This could potentially generate inappropriate results.
# Long-form voices
Amazon Polly has a **Long-form engine** that produces human-like, highly expressive, and emotionally adept voices. Long-form voices are designed to captivate listeners’ attention for longer content, such as news articles, training materials, or marketing videos.
Amazon Polly Long-form voices are developed with a cutting-edge deep learning TTS technology. The model learns to replicate phonemes, prosody, intonation, and other phonetic and acoustic aspects of human language, resulting in a highly natural speech output.
The Long-form engine uses text embeddings to interpret the meaning of a text. Using text embeddings, the Long-form engine can generate the correct emphasis, pauses, and tone of a natural voice. The result is a voice that combines the complete range of emotional elements present in human communication. This includes mimicking surprisal or differentiating dialogue from narration. Together, this creates a premium speech product that sounds like a live human being.
**Note**
The state-of-the-art technology underlying these voices falls within the paradigm of generative AI for language and voice modelling. A side effect of the technology is that any updates to the training data and the model could result in a slight variations to the way the voices sound, even in case when their overall quality improves with model updates. This could have an impact on use cases with different content parts synthesized over a long time period – for example, a season of podcasts.
## Available long-form voices
Amazon Polly currently offers four en-US and two es-ES long-form voices. Both languages have female and male voices available. The English long-form voices Daniel, Gregory, and Ruth are also available in a conversational NTTS variant.
| | Language | Language code | Name/ID | Gender |
| --- | --- | --- | --- | --- |
| 1 | **English (US)** | en-US | Danielle Gregory Ruth Patrick | Female Male Female Male |
| 2 | **Spanish (Spain)** | es-ES | Alba Raúl | Female Male |
## Feature and region compatibility
Amazon Polly long-form voices are available in the following regions:
+ US East (N. Virginia): us-east-1
+ Other regions not available
**The Amazon Polly Long-form engine supports the following features:**
+ Real-time and asynchronous speech synthesis operations.
+ All [speech marks](https://docs.aws.amazon.com/polly/latest/dg/speechmarks.html).
+ Many (but not all) SSML tags are supported by Amazon Polly. For more information about NTTS-supported SSML tags, see [Supported SSML tags](https://docs.aws.amazon.com/polly/latest/dg/supportedtags.html)
+ As with standard voices, you can choose from various sampling rates to optimize the bandwidth and audio quality for your application. Valid sampling rates for standard, long-form, and neural voices are: 8 kHz, 16 kHz, 22kHz, or 24 kHz. The default for standard voices is 22 kHz. The default for long-form and neural voices is 24 kHz. Amazon Polly supports MP3, OGG (Vorbis), and raw PCM audio stream formats.
**Note**
Long-form voices cost is specified on the [Amazon Polly pricing information page](https://aws.amazon.com/polly/pricing/).
# Neural voices
Amazon Polly has a **Neural text-to-speech (NTTS) engine** that can produce even higher quality voices than its standard voices. Standard TTS voices use concatenative synthesis. The standard engine concatenates phonemes of recorded speech, producing very natural-sounding synthesized speech. However, the inevitable variations in speech and the techniques used to segment the waveforms limits the quality of speech. The Amazon Polly NTTS engine doesn't use standard concatenative synthesis to produce speech. It has two parts:
+ A neural network — that converts a sequence of phonemes (the most basic units of language) into a sequence of *spectrograms*. (Spectograms are snapshots of the energy levels in different frequency bands.)
+ A vocoder — that converts spectrograms into a nearly continuous audio signal.
The first component of the neural TTS system is a sequence-to-sequence model. This model doesn’t create its results solely from the corresponding input but also considers how the sequence of the elements of the input work together. The model chooses the spectrograms that it outputs so that their frequency bands emphasize acoustic features that the human brain uses when processing speech.
The output of this model then passes to a neural vocoder. This converts the spectrograms into speech waveforms. When trained on the large datasets used to build general-purpose concatenative-synthesis systems, this sequence-to-sequence approach will yield higher-quality, more natural-sounding voices.
## Available neural voices
Neural voices are available in 36 languages and language variants. The following table lists the voices.
| | Language and language variants | Language code | Name/ID | Gender |
| --- | --- | --- | --- | --- |
| 1 | **Arabic (Gulf)** | ar-AE | Hala Zayd | Female Male |
| 2 | **Belgian Dutch (Flemish)** | nl-BE | Lisa | Female |
| 3 | **Catalan** | ca-ES | Arlet | Female |
| 4 | **Czech** | cs-CZ | Jitka | Female |
| 5 | **Chinese (Cantonese)** | yue-CN | Hiujin | Female |
| 6 | **Chinese (Mandarin)** | cmn-CN | Zhiyu | Female |
| 7 | **Danish** | da-DK | Sofie | Female |
| 8 | **Dutch** | nl-NL | Laura | Female |
| 9 | **English (Australian)** | en-AU | Olivia | Female |
| 10 | **English (British)** | en-GB | Amy\$1 Emma Brian Arthur | Female Female Male Male |
| 11 | **English (Indian)** | en-IN | Kajal | Female |
| 12 | **English (Irish)** | en-IE | Niamh | Female |
| 13 | **English (New Zealand)** | en-NZ | Aria | Female |
| 14 | **English (Singaporean)** | en-SG | Jasmine | Female |
| 15 | English (South African) | en-ZA | Ayanda | Female |
| 16 | **English (US)** | en-US | Danielle Gregory Ivy Joanna\$1 Kendra Kimberly Salli Joey Justin Kevin Matthew\$1 Ruth Stephen | Female Male Female (child) Female Female Female Female Male Male (child) Male (child) Male Female Male |
| 17 | **Finnish** | fi-FI | Suvi | Female |
| 18 | **French (Belgian)** | fr-BE | Isabelle | Female |
| 19 | **French (Canadian)** | fr-CA | Gabrielle Liam | Female Male |
| 20 | **French** | fr-FR | Léa Rémi | Female Male |
| 21 | **German** | de-DE | Vicki Daniel | Female Male |
| 22 | **German (Austrian)** | de-AT | Hannah | Female |
| 23 | **German (Swiss)** | de-CH | Sabrina | Female |
| 24 | **Hindi** | hi-IN | Kajal | Female |
| 25 | **Italian** | it-IT | Bianca Adriano | Female Male |
| 26 | **Japanese** | ja-JP | Takumi Kazuha Tomoko | Male Female Female |
| 27 | **Korean** | ko-KR | Seoyeon Jihye | Female Female |
| 28 | **Norwegian** | nb-NO | Ida | Female |
| 29 | **Polish** | pl-PL | Ola | Female |
| 30 | **Portuguese (Brazilian)** | pt-BR | Camila Vitória/Vitoria Thiago | Female Female Male |
| 31 | **Portuguese (European)** | pt-PT | Inês/Ines | Female |
| 32 | **Spanish (Spain)** | es-ES | Lucia Sergio | Female Male |
| 33 | **Spanish (Mexican)** | es-MX | Mia Andrés | Female Male |
| 34 | **Spanish (US)** | es-US | Lupe\$1 Pedro | Female Male |
| 35 | **Swedish** | sv-SE | Elin | Female |
| 36 | **Turkish** | tr-TR | Burcu | Female |
\$1The Amy, Joanna, Lupe, and Matthew voices can be used with the Newscaster speaking style. For more information, see [Applying the newscaster voice](newscaster-voices.md).
## Feature and region compatibility
Neural voices aren't available in all AWS Regions, nor do they support all Amazon Polly features.
Neural voices are supported in the following regions:
+ US East (N. Virginia): us-east-1
+ US West (Oregon): us-west-2
+ Africa (Cape Town): af-south-1
+ Asia Pacific (Tokyo): ap-northeast-1
+ Asia Pacific (Seoul): ap-northeast-2
+ Asia Pacific (Osaka): ap-northeast-3
+ Asia Pacific (Mumbai): ap-south-1
+ Asia Pacific (Singapore): ap-southeast-1
+ Asia Pacific (Sydney): ap-southeast-2
+ Asia Pacific (Malaysia): ap-southeast-5
+ Canada (Central): ca-central-1
+ Europe (Frankfurt): eu-central-1
+ Europe (Ireland): eu-west-1
+ Europe (London): eu-west-2
+ Europe (Paris): eu-west-3
+ Europe (Spain): eu-south-2
+ Europe (Zurich): eu-central-2
+ AWS GovCloud (US-West): us-gov-west-1
Endpoints and protocols for these Regions are identical to those used for standard voices. For more information, see [Amazon Polly endpoints and quotas](https://docs.aws.amazon.com/general/latest/gr/pol.html).
The following features are supported for neural voices:
+ Real-time and asynchronous speech synthesis operations.
+ Newscaster speaking style. For more information about the speaking styles, see [Applying the newscaster voice](newscaster-voices.md).
+ All speech marks.
+ Many (but not all) of the SSML tags that are supported by Amazon Polly. For more information about NTTS-supported SSML tags, see Supported Tags.
As with standard voices, you can choose from various sampling rates to optimize the bandwidth and audio quality for your application. Valid sampling rates for standard and neural voices are 8 kHz, 16 kHz, 22 kHz, or 24 kHz. The default for standard voices is 22 kHz. The default for neural voices is 24 kHz. Amazon Polly supports MP3, OGG (Vorbis), and raw PCM audio stream formats.
# Standard voices
Amazon Polly has a **standard** engine that use concatenative synthesis. The standard engine concatenates phonemes of recorded speech, producing very natural-sounding synthesized speech.
## Available Standard voices
Amazon Polly currently offers 40 female and 20 male standard voices in 29 language and language variants.
| | Language | Language code | Name/ID | Gender |
| --- | --- | --- | --- | --- |
| 1 | **Arabic** | arb | Zeina | Female |
| 2 | **Chinese (Mandarin)** | cmn-CN | Zhiyu | Female |
| 3 | **Danish** | da-DK | Naja Mads | Female Male |
| 4 | **Dutch** | nl-NL | Lotte Ruben | Female Male |
| 5 | **English (Australian)** | en-AU | Nicole Russell | Female Male |
| 6 | **English (British)** | en-GB | Amy Emma Brian | Female Female Male |
| 7 | **English (Indian)** | en-IN | Aditi Raveena | Female Female |
| 8 | **English (US)** | en-US | Ivy Joanna Kendra Kimberly Salli Joey Kevin | Female Female Female Female Female Male Male |
| 9 | **English (Welsh)** | en-GB-WLS | Geraint | Male |
| 10 | **French** | fr-FR | Céline/Celine Léa Mathieu | Female Female Male |
| 11 | **French (Canadian)** | fr-CA | Chantal | Female |
| 12 | **German** | de-DE | Marlene Vicki Hans | Female Female Male |
| 13 | **Hindi** | hi-IN | Aditi | Female |
| 14 | **Icelandic** | is-IS | Dóra/Dora Karl | Female Male |
| 15 | **Italian** | it-IT | Carla Bianca Giorgio | Female Female Male |
| 16 | **Japanese** | ja-JP | Mizuki Takumi | Female Male |
| 17 | **Korean** | ko-KR | Seoyeon | Female |
| 18 | **Norwegian** | nb-NO | Liv | Female |
| 19 | **Polish** | pl-PL | Ewa Maja Jacek Jan | Female Female Male Male |
| 20 | **Portuguese (Brazilian)** | pt-BR | Camila Vitória/Vitoria Ricardo | Female Female Male |
| 21 | **Portuguese (European)** | pt-PT | Inês/Ines Cristiano | Female Male |
| 22 | **Romanian** | ro-RO | Carmen | Female |
| 23 | **Russian** | ru-RU | Tatyana Maxim | Female Male |
| 24 | **Spanish (Spain)** | es-ES | Conchita Lucia Enrique | Female Female Male |
| 25 | **Spanish (Mexican)** | es-MX | Mia | Female |
| 26 | **Spanish (US)** | es-US | Lupe Penélope/Penelope Miguel | Female Female Male |
| 27 | **Swedish** | sv-SE | Astrid | Female |
| 28 | **Turkish** | tr-TR | Filiz | Male |
| 29 | **Welsh** | cy-GB | Gwyneth | Female |
## Feature and region compatibility
Amazon Polly standard voices are available in the following Amazon Polly regions:
+ US East (N. Virginia): us-east-1
+ US East (Ohio): us-east-2
+ US West (N. California): us-west-1
+ US West (Oregon): us-west-2
+ Africa (Cape Town): af-south-1
+ Asia Pacific (Hong Kong): ap-east-1
+ Asia Pacific (Tokyo): ap-northeast-1
+ Asia Pacific (Seoul): ap-northeast-2
+ Asia Pacific (Osaka): ap-northeast-3
+ Asia Pacific (Mumbai): ap-south-1
+ Asia Pacific (Singapore): ap-southeast-1
+ Asia Pacific (Sydney): ap-southeast-2
+ Asia Pacific (Malaysia): ap-southeast-5
+ China (Ningxia): cn-northwest-1;
+ Canada (Central): ca-central-1
+ Europe (Frankfurt): eu-central-1
+ Europe (Ireland): eu-west-1
+ Europe (London): eu-west-2
+ Europe (Paris): eu-west-3
+ Europe (Spain): eu-south-2
+ Europe (Stockholm): eu-north-1
+ Middle East (Bahrain): me-south-1
+ South America (São Paulo): sa-east-1
+ AWS GovCloud (US-West): us-gov-west-1
Endpoints and protocols for these Regions are identical to those used for Neural voices. For more information, see [Amazon Polly endpoints and quotas](https://docs.aws.amazon.com/general/latest/gr/pol.html).
**The Amazon Polly standard engine supports the following features (TBD):**
+ Real-time and asynchronous speech synthesis operations.
+ All [speech marks](https://docs.aws.amazon.com/polly/latest/dg/speechmarks.html).
+ Many (but not all) SSML tags are supported by Amazon Polly. For more information about NTTS-supported SSML tags, see [Supported SSML tags](https://docs.aws.amazon.com/polly/latest/dg/supportedtags.html).
+ You can choose from various sampling rates to optimize the bandwidth and audio quality for your application. The default sampling rates for standard voices are 22 kHz. Amazon Polly supports MP3, OGG (Vorbis), and raw PCM audio stream formats.
**Note**
Standard voices cost is specified on the [Amazon Polly pricing information page](https://aws.amazon.com/polly/pricing/).
# Choosing a voice engine
You can access Amazon Polly voices through the Amazon Polly console or AWS CLI.
**To choose a voice engine on the console**
1. Open the Amazon Polly console at [https://console.aws.amazon.com/polly/](https://console.aws.amazon.com/polly/).
1. From the Amazon Polly console, choose the desired voice engine.
1. Choose the desired voice from the voice drop-down menu.
1. Generate TTS audio with text of your choice.
To choose a voice engine in the AWS CLI, specify the `Engine` and `VoiceId` in the `SyntheszieSpeech` or `StartSpeechSynthesisTask` API operations. For some examples, see the [quick-start code samples](https://docs.aws.amazon.com/polly/latest/dg/get-started-what-next.html) and the [Python examples](https://docs.aws.amazon.com/polly/latest/dg/get-started-what-next.html).