@totemheart/bridge-tts

Maps Totemheart's real ExpressionDirectives.getProsodyDirectives() output ({ pitchShift, rateShift, energyLevel, breathiness }) onto standard W3C SSML <prosody> markup — the real published standard Azure Cognitive Services Speech, Amazon Polly, and Google Cloud Text-to-Speech all accept as-is, chosen over any single vendor's proprietary API so this one mapping stays honestly reusable.

This package makes no network calls and bundles no vendor SDK — it only produces a markup string. Sending it to a real TTS engine (with your own real credentials) is the caller's job.

Install

npm install totemheart @totemheart/bridge-tts

Usage

import { Totemheart, Personality } from 'totemheart'
import { TTSBridge }                from '@totemheart/bridge-tts'

const ai       = new Totemheart( { personality: new Personality() } )
const bridge = new TTSBridge()

const result = await ai.processInput( 'estoy muy emocionado por esto', { userId: 'u' } )
const prosody = ai.expressionDirectives.getProsodyDirectives( ai.emotionSpace.vector )
const ssml       = bridge.toSSML( prosody, result.text )

// ssml is now real, valid SSML — feed it directly to Azure/Polly/Google's synthesize call

API

toSSMLAttributes(prosody)

Returns { rate, pitch, volume } as real SSML-style relative percent strings (e.g. '+20%', '-10%'), derived directly from the real pitchShift/rateShift/energyLevel numbers ExpressionDirectives produces.

toSSML(prosody, text)

Wraps real text in a complete <speak>/<prosody> SSML document, XML-escaping the text. Adds a real <break time="120ms"/> when breathiness > 0.2.

License

MIT