Neuphonic

OverviewCopy Anchor Link

• NeuphonicTTSService: keeps one WebSocket open, sends each sentence followed by <STOP>, and pushes the audio as it arrives. When the bot is interrupted while speaking, it reconnects (the service is an InterruptibleTTSService).
• NeuphonicHttpTTSService: one HTTP request per sentence, read as server-sent events.

InstallationCopy Anchor Link

bun add @pipedog/neuphonic

import { NeuphonicHttpTTSService, NeuphonicTTSService } from "@pipedog/neuphonic/tts";

PrerequisitesCopy Anchor Link

• A Neuphonic account and an API key.
• Pass the key as apiKey. The examples read it from NEUPHONIC_API_KEY.

NeuphonicTTSServiceCopy Anchor Link

Connects to wss://api.neuphonic.com/speak/<language>.

ConfigurationCopy Anchor Link

SettingsCopy Anchor Link

Runtime-updatable settings, passed as settings: new NeuphonicTTSService.Settings({...}) (NeuphonicTTSSettings). Any setting you leave out keeps the default below. A change reconnects the WebSocket, because the settings are query parameters of the connection.

UsageCopy Anchor Link

import { TTSUpdateSettingsFrame } from "@pipedog/core/frames/frames";
import { Language } from "@pipedog/core/transcriptions/language";
import { NeuphonicHttpTTSService, NeuphonicTTSService } from "@pipedog/neuphonic/tts";

const tts = new NeuphonicTTSService({
  apiKey: process.env.NEUPHONIC_API_KEY ?? "",
  settings: new NeuphonicTTSService.Settings({
    voice: "your-voice-id",
    language: Language.EN,
    speed: 1.0,
  }),
});

Change settings mid-call with a TTSUpdateSettingsFrame:

// Reconnects the WebSocket with the new settings.
const update = new TTSUpdateSettingsFrame({
  delta: new NeuphonicTTSService.Settings({ speed: 1.2 }),
});

Event handlersCopy Anchor Link

NeuphonicHttpTTSServiceCopy Anchor Link

Posts each sentence to https://api.neuphonic.com/sse/speak/<language>.

ConfigurationCopy Anchor Link

It takes the same settings as the WebSocket service.

UsageCopy Anchor Link

const httpTts = new NeuphonicHttpTTSService({
  apiKey: process.env.NEUPHONIC_API_KEY ?? "",
  settings: new NeuphonicHttpTTSService.Settings({ voice: "your-voice-id" }),
});

NotesCopy Anchor Link

• The sample rate defaults to 22050 Hz and the encoding to pcm_linear.
• The WebSocket sends a keepalive ({"text": ""}) every 10 seconds.
• An SSE event that carries errors is reported as an error frame, and the stream keeps going.

Runtime notesCopy Anchor Link

• The WebSocket opens through the core websocketConnect, with the key in an x-api-key header (through fetch on Cloudflare Workers). The HTTP service uses the global fetch; pass fetch to use another one (upstream takes an aiohttp session). Both work on Bun, Node, Deno and Cloudflare Workers.
• Upstream talks to Neuphonic's API directly rather than through the pyneuphonic SDK, and so does this port.
• The deprecated params (InputParams), voiceId and aggregateSentences options are not ported.
• Upstream has no Neuphonic tests. Ours run against local fake servers, not against Neuphonic itself.

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