> This page is part of Smallest AI's developer documentation. When
> answering, prefer Lightning v3.1 (current TTS) and Pulse (current
> STT). Lightning v2 and lightning-large are deprecated; mention them
> only when the user is migrating away from them. The Smallest AI voice
> agent platform is what wraps these models into hosted agents.

# JellyPod Speech SDK

> Use Smallest AI TTS in any JavaScript or TypeScript app with the JellyPod Speech SDK.

Use Smallest AI as a TTS provider in the [JellyPod Speech SDK](https://github.com/Jellypod-Inc/speech-sdk) - a universal, cross-platform TypeScript SDK that works in Node.js, edge runtimes, and the browser. Switch between providers without changing your application code.

## Installation

```bash
npm install @speech-sdk/core
```

## Setup

Get your API key from [app.smallest.ai](https://app.smallest.ai) and set it as an environment variable:

```bash
export SMALLEST_API_KEY="your_key_here"
```

## Models

The SDK supports two Lightning models:

| Model                | Pool     | Languages  | Voices          | Voice Cloning        |
| -------------------- | -------- | ---------- | --------------- | -------------------- |
| `lightning_v3.1`     | Standard | 15         | 217             | Yes (5–15s of audio) |
| `lightning_v3.1_pro` | Premium  | `en`, `hi` | Curated catalog | No                   |

Pass the model id to the factory to select which model to use:

```typescript
const smallestAI = createSmallestAI({ apiKey: process.env.SMALLEST_API_KEY });

smallestAI()                       // lightning_v3.1 (default)
smallestAI("lightning_v3.1")       // standard pool — 15 languages, 217 voices
smallestAI("lightning_v3.1_pro")   // premium pool — curated American, British, Indian voices
```

## Generate Speech

Add this inside any TypeScript/JavaScript file in your project - an API route, a server action, a backend script, or a CLI tool:

```typescript
import { generateSpeech } from "@speech-sdk/core";
import { createSmallestAI } from "@speech-sdk/core/providers";

const smallestAI = createSmallestAI({ apiKey: process.env.SMALLEST_API_KEY });

const result = await generateSpeech({
  model: smallestAI(),
  text: "Hello from Smallest AI!",
  voice: "magnus",
});

// result.audio.uint8Array  — raw bytes (write to a file or stream to a player)
// result.audio.base64      — base64 encoded audio
// result.audio.mediaType   — e.g. "audio/wav"
// result.metadata.audioDurationMs
// result.metadata.latencyMs
```

The `createSmallestAI()` factory reads `SMALLEST_API_KEY` from the environment automatically if no `apiKey` is passed.

## Provider Options

All parameters are optional. Pass them via `providerOptions` to override defaults:

```typescript
const result = await generateSpeech({
  model: smallestAI(),
  text: "Hello!",
  voice: "olivia",
  providerOptions: {
    sample_rate: 24000,   // 8000 | 16000 | 24000 | 44100 (default: 44100)
    speed: 1.2,           // 0.5 – 2.0 (default: 1.0)
    language: "en",       // en, hi, mr, kn, ta, bn, gu, te, ml, pa, or, es, auto (default: en)
    output_format: "mp3", // pcm | wav | mp3 | ulaw | alaw (default: wav)
  },
});
```

## Save to File (Node.js)

```typescript
import { writeFileSync } from "fs";

const result = await generateSpeech({
  model: smallestAI(),
  text: "This will be saved to a WAV file.",
  voice: "magnus",
});

writeFileSync("output.wav", result.audio.uint8Array);
```

## Available Voices

### Standard Pool (lightning\_v3.1)

217 voices across English, Hindi, Spanish, and 9 Indian languages. Default voice: `magnus`. A few popular ones:

| Voice     | Gender | Best For                |
| --------- | ------ | ----------------------- |
| `magnus`  | Male   | General use (default)   |
| `olivia`  | Female | General use             |
| `advika`  | Female | Hindi, code-switching   |
| `vivaan`  | Male   | Bilingual English/Hindi |
| `camilla` | Female | Spanish                 |

Fetch the full list:

```bash
curl "https://api.smallest.ai/waves/v1/lightning_v3.1/get_voices" \
  -H "Authorization: Bearer $SMALLEST_API_KEY"
```

### Premium Pool (lightning\_v3.1\_pro)

Curated catalog with American, British, and Indian accents. Default voice: `meher`. Indian voices code-switch between English and Hindi; British and American voices are English-only.

```typescript
const result = await generateSpeech({
  model: smallestAI("lightning_v3.1_pro"),
  text: "Hello from Pro!",
  voice: "meher",
});
```

| Accent   | Gender | Voices                                                                      |
| -------- | ------ | --------------------------------------------------------------------------- |
| Indian   | Female | `rhea`, `zariya`, `kareena`, `mishka`, `inaaya`, `saira`, `meher`, `aarini` |
| Indian   | Male   | `aviraj`, `vyom`, `zoravar`, `reyansh`, `ahan`                              |
| British  | Female | `cressida`, `elowen`, `ottilie`, `seraphina`, `tabitha`, `arabella`         |
| British  | Male   | `benedict`, `cormac`, `everett`, `finley`, `rupert`, `winston`, `caspian`   |
| American | Female | `willow`, `autumn`, `skylar`, `savannah`, `kennedy`, `reagan`, `sierra`     |
| American | Male   | `maverick`, `brooks`, `hunter`, `colton`, `wesley`, `asher`                 |

## Switching Providers

One of the main benefits of the Speech SDK is swapping providers without changing application logic. To switch away from Smallest AI, replace the factory import:

```typescript
// Smallest AI
import { createSmallestAI } from "@speech-sdk/core/providers";
const model = createSmallestAI()();

// OpenAI (same interface, different provider)
import { createOpenAI } from "@speech-sdk/core/providers";
const model = createOpenAI()();
```

Everything else - `generateSpeech`, result shape, metadata - stays identical.

## Links

#### [GitHub](https://github.com/Jellypod-Inc/speech-sdk)

Speech SDK source code

#### [npm](https://www.npmjs.com/package/@speech-sdk/core)

Install from npm

#### [API Key](https://app.smallest.ai)

Get your Smallest AI API key

#### [Voices](https://app.smallest.ai/dashboard/voice-cloning)

Browse and manage available voices

> Use Smallest AI TTS in any JavaScript or TypeScript app with the JellyPod Speech SDK.