Amazon Polly delivers a significant improve to its Generative engine

Amazon Polly Generative engine now provides extra pure and correct speech output. We’re introducing an improve to the Amazon Polly Generative engine that delivers extra expressive, extra correct speech output with low latency. The improve provides 14 new voices throughout 9 locales, together with normal availability of Generative voices in Hindi and Finnish for the primary time. The improve is now typically out there in manufacturing.

Importantly, it is a drop-in improve for present generative voices. Voice names and API calls are unchanged, so your present functions proceed to work as they do at present.

Amazon Polly is a completely managed AWS service that converts textual content into natural-sounding speech for contact facilities, giant language mannequin (LLM)-based agentic functions, media and leisure, e-learning, accessibility, and different functions that require dependable voice output at scale.

Hear the distinction

You should use pattern audio to guage the enhancements straight.

The extra pure, genuine, and expressive voices. Hear how Tiffany welcomes you in a pleasant tone.
- “Hello there! I’m Tiffany, certainly one of Polly’s generative voices. I used to be constructed to sound pure, expressive, and a bit of bit pleasant. What ought to we discuss?”
Correct speech technology for numbers, foreign money, addresses, and so forth. Kenneth reads again an order affirmation, together with the monitoring quantity, date and time, foreign money and handle precisely.
- “Thanks for calling. Your order, quantity 4471 9902 ships at present to 85 Harbor View Drive, Condominium 4B, Portland, Oregon 97209. It ought to arrive on Thursday, October twenty second, between 9 AM and 1 PM. Your complete was $1,284.50 and your affirmation code is Ok-7-Q-2-X-9.”
Internet new languages. Arjun and Kiara for Hindi, and Suvi for Finnish. Hindi and Finnish are typically out there for the primary time on the Amazon Polly Generative engine.
- Arjun: नमस्कार, आप सुन रहे हैं आज की मुख्य ख़बरें। देश के कई हिस्सों में मानसून की विदाई के साथ मौसम सुहाना हो गया है। (Whats up, you’re listening to at present’s prime tales. Because the monsoon retreats throughout a lot of the nation, the climate has turned nice.)
- Kiara: नमस्ते! मेरा नाम कियारा है। आइए, आज मैं आपको एक छोटी सी कहानी सुनाती हूँ।(Whats up! My title is Kiara. Come, let me let you know a bit of story at present.)
- Suvi: Majakanvartija ei ollut nähnyt laivaa yhteentoista päivään. Sitten, vähän puolenyön jälkeen, horisontissa välähti yksi ainoa valo. Kerran. Kahdesti. Ja sitten, uskomatonta kyllä, se vilkutti takaisin hänen nimensä. (The lighthouse keeper hadn’t seen a ship in eleven days. Then, simply after midnight, a single mild blinked on the horizon. As soon as. Twice. After which, impossibly, it blinked again his title.)
Completely different types of speech supply based mostly on textual context. Powered by the brand new Amazon Polly Generative engine, the voices can mirror emotion and prosody even higher based mostly on the textual content enter, with out handbook setup.
- Impartial: “Your package deal arrived early.”
- Uncertainty: “So, um… your package deal arrived early. I believe?”
- Shocked: “Wait, what? Your package deal arrived early?”

New voices, languages, and Area availability

A bigger catalog general for Amazon Polly Generative engine – Totaling 57 generative voices throughout 11 languages and 25 locales.
14 new voices throughout 9 locales – US English (Kenneth, Nate), Australian English (Russell, Curtis), Indian English (Kiara, Arjun), German (Tina), US Spanish (Carlos), Brazilian Portuguese (Leo, Carolina), Korean (Hyunduk), Hindi (Kiara, Arjun) and Finnish (Suvi).
Internet new languages typically out there in Generative engine – Hindi and Finnish.
Area growth to AWS GovCloud (US-West) – Totaling 11 AWS Areas.

Constructed for production-ready speech workflows for enterprise

The upgraded Amazon Polly Generative engine makes use of a normal Amazon Polly API name to construct your speech functions. With Amazon Polly, you possibly can proceed to construct with acquainted AWS patterns: name the identical Amazon Polly API, handle entry with AWS Id and Entry Administration (IAM), use the output in your present functions and contact-center flows, and monitor the workloads that synthesize speech to your clients.

This launch is designed to raise the speech output high quality that you may hear together with your present Amazon Polly implementation. No adjustments to your code are required to get the upgraded voices. For functions that already use Amazon Polly Generative voices, the improve occurs in place. For brand spanking new functions, the identical SynthesizeSpeech API provides you entry to a bigger set of generative voice catalogs. With the Bi-Di Streaming API StartSpeechSynthesisStream, you possibly can stream textual content to Amazon Polly and obtain synthesized audio again concurrently. That is well-suited for real-time functions like voice brokers and bespoke characters in video games.

Aggressive price-performance

The Amazon Polly Generative engine provides aggressive pricing on a pay-as-you-go foundation. With this mannequin improve, you get improved voice high quality and accuracy on the similar value. There’s no price enhance, no new SKU, and no adjustments to your present integration. You possibly can cache and replay synthesized static prompts with no extra price. Consult with Amazon Polly pricing for full price info.

The Amazon Polly Generative engine now delivers extra pure, expressive speech alongside low-latency streaming for real-time functions, Speech Synthesis Markup Language (SSML) assist for fine-grained management, and the operational reliability of a completely managed AWS service. This mixture of high quality, efficiency, and price effectivity helps a variety of manufacturing use circumstances.

What you possibly can construct

Contact facilities. Accuracy issues most in customer support conversations. When a caller checks an order, a voice that reads again the monitoring quantity, handle, and affirmation code accurately helps resolve the decision in self-service reasonably than escalating to a human agent. Amazon Polly Generative voice’s correct entity pronunciation and low-latency technology make automated interactive voice response (IVR) interactions sound extra conversational.

Cascaded voice brokers. Utilizing Amazon Polly in a voice agent framework, you possibly can convert LLM response textual content into clear, natural-sounding speech and stream it again to customers in close to actual time over WebSockets or telephony channels. This makes reside conversational voice doable. The first advantages are low-latency playback, excessive scalability, and price effectivity.

Media, localization, and dubbing. Convert scripts into completed audio for video, promoting, podcasts, product overviews, and audiobooks. You should use one voice or many, and not using a studio session. Extra pure intonation reduces post-production. The expanded locale protection, together with first-time generative Hindi and Finnish, means you possibly can localize a single piece of content material into extra markets from the identical pipeline.

E-learning and coaching. Narrate courseware, onboarding, and knowledge-based articles at scale, and replace them as quickly because the supply textual content adjustments. No re-recording is required. Extra expressive supply helps engagement by long-form materials.

Accessibility. Present a transparent, pure spoken model of articles, paperwork, and functions so extra individuals can eat your content material by listening: learn aloud, on the go, or with a display screen reader.

Interactive and transactional experiences. Voice-enable video games, kiosks, IVR menus, and IoT gadgets with responses generated on demand. Use Amazon Polly when the voice must sound pure and browse names, numbers, product particulars, addresses, and confirmations reliably.

“At Sanoma Media Finland, we use Amazon Polly to show our real-time information articles into audio, giving the listener the selection to hear at any time when it fits them higher. Right now we publish over 10,000 articles per thirty days utilizing Polly’s Finnish neural voice. Based mostly on our testing, Suvi’s new generative voice is a transparent enchancment over the earlier neural voice – it sounds noticeably extra pure and makes for a extra fulfilling expertise for the listener.”

— Justus Reinikainen, Product Supervisor, Sanoma Media Finland

Getting began

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

Conditions:

Energetic AWS account with Amazon Polly entry.
AWS Command Line Interface (AWS CLI) put in and configured.
Python 3.10+.
Boto3 put in: pip set up boto3.
AWS Id and Entry Administration (IAM) permissions: polly:SynthesizeSpeech.

Right here’s a fast instance utilizing the AWS SDK for Python (Boto3) with the SynthesizeSpeech API and a generative voice:

import boto3

polly = boto3.consumer("polly", region_name="us-east-1")

response = polly.synthesize_speech(
    Textual content="Whats up! That is Amazon Polly's next-generation generative voice.",
    VoiceId="Kenneth",  # any generative voice within the catalogue
    Engine="generative",
    OutputFormat="mp3",
)

with open("speech.mp3", "wb") as f:
    f.write(response["AudioStream"].learn())

You can too make the identical request with the AWS CLI:

aws polly synthesize-speech 
  --region us-east-1 
  --voice-id Kenneth 
  --engine generative 
  --output-format mp3 
  --text "Whats up! That is Amazon Polly's next-generation generative voice." 
  speech.mp3

To retrieve the complete, present voice checklist programmatically, name DescribeVoices. You should use this in deployment checks to verify {that a} chosen generative voice is on the market within the Area the place your software runs.

Availability

The following-generation Generative engine and the brand new voices at the moment are typically out there. The 14 new Generative voices can be found within the following AWS Areas: US East (N. Virginia), US West (Oregon), Europe (Frankfurt), Europe (Zurich), Europe (London), Asia Pacific (Sydney), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Tokyo), Canada (Central), and AWS GovCloud (US-West). Getting began, by viewers:

Builders – Name synthesize_speech with Engine="generative" in your present software. No different adjustments are required.
Contact-center groups – The enhancements attain Amazon Join and Genesys flows routinely.
Everybody – Entry any voice within the Amazon Polly console.

Just a few issues to know:

Normal and neural voices are unaffected.
Speech marks aren’t out there on generative voices (unchanged from at present). In case your software requires them, commonplace and neural voices proceed to assist them.
Generative engine continues to assist SSML tags. See supported SSML tags for extra particulars.
Use present voice availability at runtime. Name DescribeVoices or verify the console when deploying throughout Areas.

Conclusion

The following-generation Generative engine of Amazon Polly is on the market at present, delivering extra pure and extra correct voices, including 14 new voices, together with generative voices typically out there in Hindi and Finnish for the primary time, and requires no change to your code. See Amazon Polly pricing web page for pricing info. To get began, open the Amazon Polly console to attempt the voices, and see the Amazon Polly documentation to construct with them.