AssemblyAI

The AssemblyAISTTService provides real-time speech-to-text transcription using AssemblyAI's streaming WebSocket API. It's optimized for high accuracy and supports features like transcript formatting and end-of-turn detection.

Installation

To use AssemblyAI, install the required dependencies:

pip install "piopiy-ai[assemblyai]"

Prerequisites

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

  export ASSEMBLYAI_API_KEY="your_api_key_here"

Configuration

AssemblyAISTTService Parameters

Parameter	Type	Default	Description
api_key	str	Required	Your AssemblyAI API key.
language	Language	Language.EN	Transcription language.
api_endpoint_base_url	str	wss://streaming.assemblyai.com/v3/ws	WebSocket endpoint URL.
connection_params	AssemblyAIConnectionParams	AssemblyAIConnectionParams()	Advanced connection settings.
vad_force_turn_endpoint	bool	True	Force end-of-turn on VAD stop.

AssemblyAIConnectionParams

Parameter	Type	Default	Description
sample_rate	int	16000	Audio sample rate in Hz.
encoding	str	"pcm_s16le"	Audio encoding format (pcm_s16le, pcm_mulaw).
formatted_finals	bool	True	Enable transcript formatting (punctuation, casing).
speech_model	str	"universal-streaming-english"	Transcription model to use.

Usage

Basic Setup

import os
from piopiy.services.assemblyai.stt import AssemblyAISTTService

stt = AssemblyAISTTService(
    api_key=os.getenv("ASSEMBLYAI_API_KEY"),
)

With Custom Parameters

from piopiy.services.assemblyai.stt import AssemblyAISTTService
from piopiy.services.assemblyai.models import AssemblyAIConnectionParams

params = AssemblyAIConnectionParams(
    sample_rate=16000,
    formatted_finals=True,
    speech_model="universal-streaming-english"
)

stt = AssemblyAISTTService(
    api_key=os.getenv("ASSEMBLYAI_API_KEY"),
    connection_params=params
)

Notes

• Language Support: Currently, AssemblyAI's streaming API is optimized for English.
• Barge-in: Works seamlessly with Piopiy's built-in VAD for interruptible conversations.