Cartesia STT

The CartesiaSTTService provides cutting-edge, low-latency speech recognition using Cartesia's Sonic intelligence platform. It is optimized for deeply conversational AI flows where speed and accuracy are paramount.

Installation

To use Cartesia, install the required dependencies:

pip install "piopiy-ai[cartesia]"

Prerequisites

• A Cartesia account and API key (Get yours here).
• Set your API key in your environment:

  export CARTESIA_API_KEY="your_api_key_here"

Configuration

CartesiaSTTService Parameters

Parameter	Type	Default	Description
api_key	str	Required	Your Cartesia API key.
base_url	str	"api.cartesia.ai"	Custom Cartesia API base URL.
sample_rate	int	16000	Audio sample rate in Hz.
live_options	CartesiaLiveOptions	None	Cartesia-specific transcription options.

CartesiaLiveOptions Parameters

Option	Type	Default	Description
model	str	"ink-whisper"	The transcription model to use.
language	str	"en"	Target language for transcription.
encoding	str	"pcm_s16le"	Audio encoding format.

Usage

Basic Setup

import os
from piopiy.services.cartesia.stt import CartesiaSTTService

stt = CartesiaSTTService(
    api_key=os.getenv("CARTESIA_API_KEY")
)

With Custom Options

import os
from piopiy.services.cartesia.stt import CartesiaSTTService, CartesiaLiveOptions

options = CartesiaLiveOptions(
    model="ink-whisper",
    language="en-US"
)

stt = CartesiaSTTService(
    api_key=os.getenv("CARTESIA_API_KEY"),
    live_options=options
)

Notes

• Model Choice: ink-whisper is the default model, providing a great balance of speed and robustness.
• WebSocket Streaming: This service uses Cartesia's WebSocket API for real-time, bi-directional transcription streaming.