phonic

Phonic Node.js SDK

Node.js library for the Phonic API.

• Installation
• Setup
• Usage
  • Get voices
  • Get voice by id
  • Text-to-speech via WebSocket

Installation

npm i phonic

Setup

Grab an API key from Phonic settings and pass it to the Phonic constructor.

import { Phonic } from "phonic";

const phonic = new Phonic("ph_...");

Usage

Get voices

const { data, error } = await phonic.voices.list();

if (error === null) {
  console.log(data.voices);
}

Get voice by id

const { data, error } = await phonic.voices.get("australian-man");

if (error === null) {
  console.log(data.voice);
}

Text-to-speech via WebSocket

Open a WebSocket connection:

const { data, error } = await phonic.tts.websocket({
  model: "shasta",
  output_format: "mulaw_8000",
  voice_id: "australian-man",
});

if (error !== null) {
  throw new Error(error.message);
}

// Here we know that the WebSocket connection is open.
const { phonicWebSocket } = data;

Process audio chunks that Phonic sends back to you, by sending them to Twilio, for example:

phonicWebSocket.onMessage((message) => {
  if (message.type === "audio_chunk") {
    ws.send(
      JSON.stringify({
        event: "media",
        streamSid: "...",
        media: {
          payload: message.audio,
        },
      }),
    );
  }
});

Send text chunks to Phonic for audio generation as you receive them from LLM:

const stream = await openai.chat.completions.create(...);

for await (const chunk of stream) {
  const text = chunk.choices[0]?.delta?.content || "";

  if (text) {
    phonicWebSocket.generate({ text });
  }
}

Tell Phonic to finish generating audio for all text chunks you've sent:

phonicWebSocket.flush();

You can also tell Phonic to stop sending audio chunks back, e.g. if the user interrupts the conversation:

phonicWebSocket.stop();

To close the WebSocket connection:

phonicWebSocket.close();

To know when the last audio chunk has been received:

phonicWebSocket.onMessage((message) => {
  if (message.type === "flushed") {
    console.log("Last audio chunk received");
  }
});

You can also listen for close and error events:

phonicWebSocket.onClose((event) => {
  console.log(
    `Phonic WebSocket closed with code ${event.code} and reason "${event.reason}"`,
  );
});

phonicWebSocket.onError((event) => {
  console.log(`Error from Phonic WebSocket: ${event.message}`);
});

Release a new version to npm

1. bunx changeset
2. git add .
3. git commit -m "Add changeset"
4. git push

Git action will run and create a PR. Once this PR is merged, the new version will be released to npm.

License

MIT