Cartesia TypeScript Library

The Cartesia TypeScript library provides convenient access to the Cartesia APIs from TypeScript.

Installation

npm i -s @cartesia/cartesia-js

Reference

A full reference for this library is available here.

Usage

Instantiate and use the client with the following:

import { CartesiaClient } from "@cartesia/cartesia-js";

const client = new CartesiaClient({ apiKey: "YOUR_API_KEY" });
await client.auth.accessToken({
    grants: {
        stt: true,
    },
    expiresIn: 60,
});

Speech-to-Text (STT)

import { CartesiaClient } from "@cartesia/cartesia-js";
import fs from "node:fs";

async function streamingSTTExample() {
    const client = new CartesiaClient({
        apiKey: process.env.CARTESIA_API_KEY,
    });

    // Create websocket connection with endpointing parameters
    const sttWs = client.stt.websocket({
        model: "ink-whisper",
        language: "en", // Language of your audio
        encoding: "pcm_s16le", // Audio encoding format (required)
        sampleRate: 16000, // Audio sample rate (required)
        minVolume: 0.1, // Volume threshold for voice activity detection (0.0-1.0)
        maxSilenceDurationSecs: 2.0, // Maximum silence duration before endpointing
    });

    // Concurrent audio sending
    async function sendAudio() {
        try {
            const audioBuffer = fs.readFileSync("audio.wav");
            const chunkSize = 3200; // ~200ms chunks for more realistic streaming

            console.log("Starting audio stream...");

            for (let i = 0; i < audioBuffer.length; i += chunkSize) {
                const chunk = audioBuffer.subarray(i, i + chunkSize);
                const arrayBuffer = chunk.buffer.slice(chunk.byteOffset, chunk.byteOffset + chunk.byteLength);

                await sttWs.send(arrayBuffer);
                console.log(`Sent chunk ${Math.floor(i / chunkSize) + 1}`);

                // Simulate real-time audio capture delay
                await new Promise((resolve) => setTimeout(resolve, 100));
            }

            await sttWs.finalize();
            console.log("Audio streaming completed");
        } catch (error) {
            console.error("Error sending audio:", error);
        }
    }

    // Concurrent transcript receiving with word-level timestamps
    async function receiveTranscripts(): Promise<string> {
        return new Promise((resolve) => {
            let fullTranscript = "";

            sttWs.onMessage((result) => {
                if (result.type === "transcript") {
                    const status = result.isFinal ? "FINAL" : "INTERIM";
                    console.log(`[${status}] "${result.text}"`);

                    // Handle word-level timestamps if available
                    if (result.words && result.words.length > 0) {
                        console.log("Word-level timestamps:");
                        result.words.forEach((word) => {
                            console.log(`  "${word.word}": ${word.start.toFixed(2)}s - ${word.end.toFixed(2)}s`);
                        });
                    }

                    if (result.isFinal) {
                        fullTranscript += `${result.text} `;
                    }
                } else if (result.type === "flush_done") {
                    console.log("Flush completed - sending done command");
                    sttWs.done().catch(console.error);
                } else if (result.type === "done") {
                    console.log("Transcription completed");
                    resolve(fullTranscript.trim());
                } else if (result.type === "error") {
                    console.error(`Error: ${result.message}`);
                    resolve("");
                }
            });
        });
    }

    try {
        console.log("Starting STT processing...");

        // Run audio sending and transcript receiving concurrently
        const [, finalTranscript] = await Promise.all([sendAudio(), receiveTranscripts()]);

        console.log(`\nFinal transcript: ${finalTranscript}`);

        // Clean up
        sttWs.disconnect();

        return finalTranscript;
    } catch (error) {
        console.error("STT processing error:", error);
        sttWs.disconnect();
        throw error;
    }
}

// Run the example
streamingSTTExample().catch(console.error);

Request And Response Types

The SDK exports all request and response types as TypeScript interfaces. Simply import them with the following namespace:

import { Cartesia } from "@cartesia/cartesia-js";

const request: Cartesia.InfillBytesRequest = {
    ...
};

Exception Handling

When the API returns a non-success status code (4xx or 5xx response), a subclass of the following error will be thrown.

import { CartesiaError } from "@cartesia/cartesia-js";

try {
    await client.auth.accessToken(...);
} catch (err) {
    if (err instanceof CartesiaError) {
        console.log(err.statusCode);
        console.log(err.message);
        console.log(err.body);
    }
}

Binary Response

You can consume binary data from endpoints using the BinaryResponse type which lets you choose how to consume the data:

const response = await client.agents.downloadCallAudio(...);
const stream: ReadableStream<Uint8Array> = response.stream();
// const arrayBuffer: ArrayBuffer = await response.arrayBuffer();
// const blob: Blob = response.blob();
// const bytes: Uint8Array = response.bytes();
// You can only use the response body once, so you must choose one of the above methods.
// If you want to check if the response body has been used, you can use the following property.
const bodyUsed = response.bodyUsed;

import { createWriteStream } from 'fs';
import { Readable } from 'stream';
import { pipeline } from 'stream/promises';

const response = await client.agents.downloadCallAudio(...);

const stream = response.stream();
const nodeStream = Readable.fromWeb(stream);
const writeStream = createWriteStream('path/to/file');

await pipeline(nodeStream, writeStream);

import { writeFile } from 'fs/promises';

const response = await client.agents.downloadCallAudio(...);

const arrayBuffer = await response.arrayBuffer();
await writeFile('path/to/file', Buffer.from(arrayBuffer));

import { writeFile } from 'fs/promises';

const response = await client.agents.downloadCallAudio(...);

const blob = await response.blob();
const arrayBuffer = await blob.arrayBuffer();
await writeFile('output.bin', Buffer.from(arrayBuffer));

import { writeFile } from 'fs/promises';

const response = await client.agents.downloadCallAudio(...);

const bytes = await response.bytes();
await writeFile('path/to/file', bytes);

const response = await client.agents.downloadCallAudio(...);

const stream = response.stream();
await Bun.write('path/to/file', stream);

const response = await client.agents.downloadCallAudio(...);

const arrayBuffer = await response.arrayBuffer();
await Bun.write('path/to/file', arrayBuffer);

const response = await client.agents.downloadCallAudio(...);

const blob = await response.blob();
await Bun.write('path/to/file', blob);

const response = await client.agents.downloadCallAudio(...);

const bytes = await response.bytes();
await Bun.write('path/to/file', bytes);

const response = await client.agents.downloadCallAudio(...);

const stream = response.stream();
const file = await Deno.open('path/to/file', { write: true, create: true });
await stream.pipeTo(file.writable);

const response = await client.agents.downloadCallAudio(...);

const arrayBuffer = await response.arrayBuffer();
await Deno.writeFile('path/to/file', new Uint8Array(arrayBuffer));

const response = await client.agents.downloadCallAudio(...);

const blob = await response.blob();
const arrayBuffer = await blob.arrayBuffer();
await Deno.writeFile('path/to/file', new Uint8Array(arrayBuffer));

const response = await client.agents.downloadCallAudio(...);

const bytes = await response.bytes();
await Deno.writeFile('path/to/file', bytes);

const response = await client.agents.downloadCallAudio(...);

const blob = await response.blob();
const url = URL.createObjectURL(blob);

// trigger download
const a = document.createElement('a');
a.href = url;
a.download = 'filename';
a.click();
URL.revokeObjectURL(url);

const response = await client.agents.downloadCallAudio(...);

const stream = response.stream();
const reader = stream.getReader();
const chunks = [];

while (true) {
  const { done, value } = await reader.read();
  if (done) break;
  chunks.push(value);
}

const blob = new Blob(chunks);
const url = URL.createObjectURL(blob);

// trigger download
const a = document.createElement('a');
a.href = url;
a.download = 'filename';
a.click();
URL.revokeObjectURL(url);

const response = await client.agents.downloadCallAudio(...);

const arrayBuffer = await response.arrayBuffer();
const blob = new Blob([arrayBuffer]);
const url = URL.createObjectURL(blob);

// trigger download
const a = document.createElement('a');
a.href = url;
a.download = 'filename';
a.click();
URL.revokeObjectURL(url);

const response = await client.agents.downloadCallAudio(...);

const bytes = await response.bytes();
const blob = new Blob([bytes]);
const url = URL.createObjectURL(blob);

// trigger download
const a = document.createElement('a');
a.href = url;
a.download = 'filename';
a.click();
URL.revokeObjectURL(url);

const response = await client.agents.downloadCallAudio(...);

const stream = response.stream();
const text = await new Response(stream).text();

const response = await client.agents.downloadCallAudio(...);

const arrayBuffer = await response.arrayBuffer();
const text = new TextDecoder().decode(arrayBuffer);

const response = await client.agents.downloadCallAudio(...);

const blob = await response.blob();
const text = await blob.text();

const response = await client.agents.downloadCallAudio(...);

const bytes = await response.bytes();
const text = new TextDecoder().decode(bytes);

Pagination

List endpoints are paginated. The SDK provides an iterator so that you can simply loop over the items:

import { CartesiaClient } from "@cartesia/cartesia-js";

const client = new CartesiaClient({ token: "YOUR_TOKEN" });
const response = await client.agents.listCalls({
    agentId: "agent_id",
});
for await (const item of response) {
    console.log(item);
}

// Or you can manually iterate page-by-page
let page = await client.agents.listCalls({
    agentId: "agent_id",
});
while (page.hasNextPage()) {
    page = page.getNextPage();
}

Advanced

Additional Headers

If you would like to send additional headers as part of the request, use the headers request option.

const response = await client.auth.accessToken(..., {
    headers: {
        'X-Custom-Header': 'custom value'
    }
});

Retries

The SDK is instrumented with automatic retries with exponential backoff. A request will be retried as long as the request is deemed retriable and the number of retry attempts has not grown larger than the configured retry limit (default: 2).

A request is deemed retriable when any of the following HTTP status codes is returned:

• 408 (Timeout)
• 429 (Too Many Requests)
• 5XX (Internal Server Errors)

Use the maxRetries request option to configure this behavior.

const response = await client.auth.accessToken(..., {
    maxRetries: 0 // override maxRetries at the request level
});

Timeouts

The SDK defaults to a 60 second timeout. Use the timeoutInSeconds option to configure this behavior.

const response = await client.auth.accessToken(..., {
    timeoutInSeconds: 30 // override timeout to 30s
});

Aborting Requests

The SDK allows users to abort requests at any point by passing in an abort signal.

const controller = new AbortController();
const response = await client.auth.accessToken(..., {
    abortSignal: controller.signal
});
controller.abort(); // aborts the request

Runtime Compatibility

The SDK defaults to node-fetch but will use the global fetch client if present. The SDK works in the following runtimes:

• Node.js 18+
• Vercel
• Cloudflare Workers
• Deno v1.25+
• Bun 1.0+
• React Native

Customizing Fetch Client

The SDK provides a way for your to customize the underlying HTTP client / Fetch function. If you're running in an unsupported environment, this provides a way for you to break glass and ensure the SDK works.

import { CartesiaClient } from "@cartesia/cartesia-js";

const client = new CartesiaClient({
    ...
    fetcher: // provide your implementation here
});

Contributing

While we value open-source contributions to this SDK, this library is generated programmatically. Additions made directly to this library would have to be moved over to our generation code, otherwise they would be overwritten upon the next generated release. Feel free to open a PR as a proof of concept, but know that we will not be able to merge it as-is. We suggest opening an issue first to discuss with us!

On the other hand, contributions to the README are always very welcome!

Documentation

API reference documentation is available here.