A brand new generative engine and three voices at the moment are usually accessible on Amazon Polly

At the moment, we’re saying the final availability of the generative engine of Amazon Polly with three voices: Ruth and Matthew in American English and Amy in British English. The brand new generative engine was educated with publicly accessible and proprietary knowledge, a wide range of voices, languages, and types. It performs with the best precision to render context-dependent prosody, pausing, spelling, dialectal properties, overseas phrase pronunciation, and extra.

Amazon Polly is a machine studying (ML) service that converts textual content to lifelike speech, referred to as text-to-speech (TTS) know-how. Now, Amazon Polly consists of high-quality, natural-sounding human-like voices in dozens of languages, so you possibly can choose the perfect voice and distribute your speech-enabled functions in lots of locales or nations.

With Amazon Polly, you possibly can choose varied voice choices, together with neural, long-form, and generative voices, which ship ground-breaking enhancements in speech high quality and produce human-like, extremely expressive, and emotionally adept voices. You possibly can retailer speech output in normal codecs like MP3 or OGG, alter the speech charge, pitch, or quantity with Speech Synthesis Markup Language (SSML) tags, and rapidly ship lifelike voices and conversational consumer experiences with constantly quick response occasions.

What’s the brand new generative engine?
Amazon Polly now helps 4 voice engines: normal, neural, long-form, and generative voices.

Normal TTS voices, launched in 2016 use conventional concatenative synthesis. This methodology strings collectively the phonemes of recorded speech, producing very natural-sounding synthesized speech. Nonetheless, the inevitable variations in speech and the methods used to phase the waveforms restrict the standard of speech.

Neural TTS (NTTS) voices, launched in 2019, use a sequence-to-sequence neural community that converts a sequence of phonemes into spectrograms, and a neural vocoder that converts the spectrograms right into a steady audio sign. The NTTS produces even greater high quality human-like voices than its normal voices.

Lengthy-form voices, launched in 2023, are developed with cutting-edge deep studying TTS know-how and designed to captivate listeners’ consideration for longer content material, similar to information articles, coaching supplies, or advertising movies.

In February 2024, Amazon scientists launched a brand new analysis TTS mannequin referred to as Huge Adaptive Streamable TTS with Emergent skills (BASE). With this know-how, Polly Generative engine is ready to create human-like synthetically generated voices. You should utilize these voices as a educated buyer assistant, a digital coach, or an skilled marketer.

Listed below are the brand new generative voices:

You possibly can select from these voice choices to fit your utility and use case. To be taught extra concerning the generative engine, go to Generative voices within the AWS documentation.

Get began with utilizing generative voices
You possibly can entry the brand new voices utilizing the AWS Administration Console, AWS Command Line Interface (AWS CLI), or the AWS SDKs.

To get began, go to the Amazon Polly console within the US (N. Virginia) Area and select Textual content-to-Speech menu within the left pane. If you choose the voice of Ruth or Matthew within the language of English, US or Amy in English, UK, you possibly can select Generative engine. Enter your textual content and hearken to or obtain the generated voice output.

Utilizing the CLI, you possibly can listing the voices that use the brand new generative engine:

$ aws polly describe-voices --output json --region us-east-1 
| jq -r '.Voices[] | choose(.SupportedEngines | index("generative")) | .Title'

Matthew
Amy
Ruth

Now, run the synthesize-speech CLI command to synthesize pattern textual content to an audio file (whats up.mp3) with the parameters of generative engine and a supported voice ID.

$ aws polly synthesize-speech --output-format mp3 --region us-east-1 
  --text "Howdy. That is my first generative voices!" 
  --voice-id Matthew --engine generative whats up.mp3

To be taught extra code examples utilizing AWS SDKs, go to Code and Utility Examples within the AWS documentation. You should utilize Java and Python code examples, utility examples similar to internet functions utilizing Java or Python, or iOS and Android functions.

Now accessible
The brand new generative voices of Amazon Polly at the moment are accessible at present within the US East (N. Virginia) Area. You solely pay for what you utilize primarily based on the variety of characters of textual content that you just convert to speech. To be taught extra, go to our Amazon Polly Pricing web page.

Give new generative voices a strive within the Amazon Polly console at present and ship suggestions to AWS re:Submit for Amazon Polly or via your ordinary AWS Help contacts.

— Channy