Hathora

Speech-to-text service implementations hosted on Hathora

Overview

Hathora is a hosting provider for several models for voice AI, which can be utilized under the single HathoraSTTService.

Hathora STT API Reference

Pipecat’s API methods for Hathora-hosted STT models

Example Implementation

Complete example using Hathora-hosted models

Hathora Models Documentation

Official Hathora documentation and features

Installation

To use Hathora services, install the required dependencies:

pip install "pipecat-ai[hathora]"

Prerequisites

Hathora Account Setup

Before using Hathora STT services, you need:

1. Hathora Account: Sign up at Hathora Models Console
2. API Key: Generate an API token from your Tokens page

Hathora Model Specifier

The HathoraSTTService accepts a model: str parameter which corresponds to the model you would like to use.

You can find available specifiers here

Configuration

HathoraSTTService

model · str · required

Model to use for transcription. Find available models at models.hathora.dev.

api_key · str · default: "None"

Hathora API key for authentication. Falls back to the HATHORA_API_KEY environment variable.

base_url · str · default: "https://api.models.hathora.dev/inference/v1/stt"

Base API URL for the Hathora STT service.

sample_rate · int · default: "None"

Audio sample rate in Hz. When None, uses the pipeline’s configured sample rate.

params · InputParams · default: "None"

Optional configuration parameters. See InputParams below.

InputParams

Parameter	Type	Default	Description
language	str	None	Language code, if supported by the selected model.
config	list[ConfigOption]	None	Additional model-specific configuration options. Refer to Hathora docs for supported options per model. Each ConfigOption has a name and value field.

Usage

Basic Setup

from pipecat.services.hathora import HathoraSTTService

stt = HathoraSTTService(
    model="your-model-specifier",
    api_key=os.getenv("HATHORA_API_KEY"),
)

With Language and Config Options

from pipecat.services.hathora import HathoraSTTService
from pipecat.services.hathora.utils import ConfigOption

stt = HathoraSTTService(
    model="your-model-specifier",
    api_key=os.getenv("HATHORA_API_KEY"),
    params=HathoraSTTService.InputParams(
        language="en",
        config=[
            ConfigOption(name="option_name", value="option_value"),
        ],
    ),
)

Notes

• Segmented transcription: HathoraSTTService extends SegmentedSTTService, meaning it processes complete audio segments (after VAD detects the user has stopped speaking) rather than streaming audio in real time.
• Model-specific features: Configuration options and language support vary by model. Check the Hathora documentation for details on each model’s capabilities.