Smallest AI

Smallest AI is a text-to-speech plugin built for very low latency. It synthesizes the LLM’s reply into the agent’s voice using the Lightning model family. It occupies the pipeline’s tts slot.

Setup

Set your Smallest AI API key in the worker environment. Generate a key from the Smallest AI console:

export SMALLESTAI_API_KEY=<key>

Usage

Import the plugin and pass it to the pipeline’s tts slot.

from zeroruntime.plugins import SmallestAITTS

tts = SmallestAITTS(
    model="lightning_v3.1",
    voice_id="magnus",
)

# Pipeline(tts=tts, ...)

import { SmallestAITTS } from '@zeroruntime/js-sdk/plugins';

const tts = SmallestAITTS({
  model: 'lightning-v3.1',
  voice_id: 'magnus',
});

// Pipeline({ tts, ... })

The default model is lightning_v3.1.

Parameters

Constructor parameters for SmallestAITTS. The Python and Node JS SDKs share these field names.

Parameter	Type	Default	Description
`api_key`	`str`	`None`	Smallest AI API key. Falls back to the `SMALLESTAI_API_KEY` environment variable when unset.
`voice`	`str`	`"magnus"`	Voice name (e.g. `"magnus"`, `"lauren"`, `"meher"`, `"devansh"`). Takes precedence over `voice_id`.
`voice_id`	`str`	`None`	Alias for `voice`; ignored when `voice` is also set.
`model`	`str`	`"lightning_v3.1"`	Model: `"lightning_v3.1"`, `"lightning_v3.1_pro"`, or `"lightning"`.
`sample_rate`	`int`	`24000`	Output sample rate: one of `8000`, `16000`, `24000`, `44100`.
`language`	`str`	`"en"`	ISO 639-1 language code: `"en"`, `"hi"`, `"mr"`, `"kn"`, `"ta"`, `"bn"`, `"gu"`, `"te"`, `"ml"`, `"pa"`, `"or"`, or `"es"`.
`speed`	`float`	`None`	Speaking rate multiplier relative to the voice default; `None` uses the model default.
`stream`	`bool`	`True`	When `True`, audio is returned as a stream of chunks for lower latency; `False` returns a single synchronous response.

Import paths

SDK	Import	Constructor
Python	`from zeroruntime.plugins import SmallestAITTS`	`SmallestAITTS(...)`
Node JS	`import { SmallestAITTS } from '@zeroruntime/js-sdk/plugins'`	`SmallestAITTS({ ... })`

The synthesized audio is streamed back to the caller as the final stage of the pipeline.

Plugins

Inference Gateway

Speech-to-text

LLM

Text-to-speech

Realtime models

Avatars

Turn detection

Voice activity detection

Noise cancellation

Setup

Usage

Parameters

Import paths

​Setup

​Usage

​Parameters

​Import paths

Setup

Usage

Parameters

Import paths