Built for voice agents that can't wait 600ms

Amazon: Amazon Polly Generative

amazon/amazon-polly-generative
active tts 1 provider
Playground

Amazon Polly's largest model, a billion-parameter transformer decoded by a streamable convolutional decoder, producing the most human-like conversational voices across 43 voices and 24 language codes, without speech marks or the newscaster style.

AuthorAmazon
Languages24 languages
VoicesLoading...
Response formatsMP3 · Opus · PCM
Statusactive

Providers

Different providers can host the same model. Choose one provider when you need a fixed backend, or let routing select among them.

ProviderSlugPer 1k charactersLatencyiFor TTS, latency is measured as time to first byte of audio. For STT, latency is measured as time to first token.ModesUptime
Amazon 5 params
amazon
$0.03
—
Synchronous

Supported languages

24 languages
German (Austrian) de-AT
German (Swiss standard) de-CH
German de-DE
English (Australian) en-AU
English (British) en-GB
English (Ireland) en-IE
English (Indian) en-IN
English (New Zealand) en-NZ
English (Singaporean) en-SG
English (US) en-US
English (South African) en-ZA
Spanish (Spain) es-ES
Spanish (Mexican) es-MX
Spanish (US) es-US
French (Belgian) fr-BE
French (Canadian) fr-CA
French fr-FR
Hindi hi-IN
Italian it-IT
Korean ko-KR
Dutch (Belgian) nl-BE
Dutch nl-NL
Polish pl-PL
Portuguese (Brazilian) pt-BR

Voices

Features

Text to speech

Generates spoken audio from text input.

text_to_speech
Synchronous

SynthesizeSpeech (POST /v1/speech) takes the whole text up front and returns the finished audio as the HTTP response body.

synchronous
Multilingual

43 voices across 24 language codes, with the bilingual voice Kajal also accepting its AdditionalLanguageCodes via LanguageCode.

multilingual
Fully bilingual voices

Kajal speaks Indian English (en-IN, default) and Hindi (hi-IN); select the language with LanguageCode.

bilingual_voices
Asynchronous long-text synthesis

Upstream only: StartSpeechSynthesisTask accepts up to 100,000 billed characters and writes the audio to an S3 bucket, which this platform does not expose.

async_long_text
Pronunciation lexicons

Upstream only: up to five stored PLS lexicons (LexiconNames) apply per request when their language matches the voice.

pronunciation_lexicons

Code

Client
Mode
Language

ElevenLabs SDK streaming TTS

Routes the ElevenLabs SDK through allmodels and streams the audio response.

import { ElevenLabsClient } from "elevenlabs";

const client = new ElevenLabsClient({
  apiKey: process.env.ALLMODELS_API_KEY,
  baseUrl: "https://api.allmodels.io/el"
});

const stream = await client.textToSpeech.stream("voice_id", {
  text: "I'm sorry, Dave. I'm afraid I can't do that.",
  modelId: "amazon/amazon-polly-generative",
  outputFormat: "mp3_44100_128"
}, {
  queryParams: {
    provider: {
      "only": [
        "amazon"
      ]
    }
  }
});

const reader = stream.getReader();
for (;;) {
  const { done, value } = await reader.read();
  if (done) break;
  // Write value to your response, file, or audio playback pipeline.
}