AWS Polly

OverviewCopy Anchor Link

AWSPollyTTSService synthesizes each sentence with Amazon Polly's SynthesizeSpeech API. It wraps the text in SSML (with prosody when rate, pitch or volume are set), requests 16 kHz PCM and resamples it to the pipeline's output rate.

InstallationCopy Anchor Link

bun add @pipedog/aws

import { AWSPollyTTSService } from "@pipedog/aws/tts";

PrerequisitesCopy Anchor Link

• An IAM identity allowed polly:SynthesizeSpeech.

• Keys, in this order of precedence:

  1. awsAccessKeyId, apiKey (the secret access key) and awsSessionToken.
  2. The AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY and AWS_SESSION_TOKEN environment variables.
  3. A credentialsProvider function, called for every request.

  The region comes from region, else AWS_REGION, else us-east-1.

AWSPollyTTSServiceCopy Anchor Link

ConfigurationCopy Anchor Link

SettingsCopy Anchor Link

Runtime-updatable settings, passed as settings: new AWSPollyTTSService.Settings({...}).

UsageCopy Anchor Link

import { AWSPollyTTSService } from "@pipedog/aws/tts";
import { TTSUpdateSettingsFrame } from "@pipedog/core/frames/frames";

const tts = new AWSPollyTTSService({
  // Or leave these out to read the AWS_* environment variables.
  awsAccessKeyId: process.env.AWS_ACCESS_KEY_ID,
  apiKey: process.env.AWS_SECRET_ACCESS_KEY,
  region: "us-east-1",
  sampleRate: 24000,
  settings: new AWSPollyTTSService.Settings({
    voice: "Joanna",
    engine: "generative",
    language: "en-US",
    rate: "105%",
  }),
});

Change the voice mid-call with a TTSUpdateSettingsFrame:

// Queue it on the worker, or push it from a processor upstream of the service.
const update = new TTSUpdateSettingsFrame({
  delta: new AWSPollyTTSService.Settings({ voice: "Matthew" }),
});

NotesCopy Anchor Link

• One HTTP request per sentence. The audio is pushed as TTSAudioRawFrames of chunkSize bytes once the response has arrived.
• Text is XML-escaped and sent as SSML (TextType: "ssml").
• Any failed request (an HTTP error, a network error, missing credentials) becomes an ErrorFrame. Upstream does this for botocore errors only.

Runtime notesCopy Anchor Link

Generated from pipecat.runtimes and pipecat.compat in packages/aws/package.json: ✅ means the compat smoke test passes on that runtime.

• Runs on Bun, Node, Deno and Cloudflare Workers: POST /v1/speech with fetch, signed with SigV4 on WebCrypto.
• There is no default AWS credential chain (profiles, IMDS, ECS roles, SSO). Use keys, the environment, or credentialsProvider.
• endpointUrl and fetch replace the Polly host and the global fetch.
• Not ported: the deprecated InputParams, and the voiceId and params constructor options. Use settings.