AWS Transcribe

The AWSTranscribeSTTService provides real-time speech-to-text using AWS Transcribe's streaming WebSocket API. It is a robust choice for developers already using the AWS infrastructure.

Installation

To use AWS Transcribe, install the required dependencies:

pip install "piopiy-ai[aws]"

Prerequisites

• An AWS account with permissions for transcribe:StartStreamTranscription (Get yours here).
• AWS Access Key, Secret Key, and Region.
• Configure your environment or AWS credentials file:

  export AWS_ACCESS_KEY_ID="your_access_key"
  export AWS_SECRET_ACCESS_KEY="your_secret_key"
  export AWS_REGION="us-east-1"

Configuration

AWSTranscribeSTTService Parameters

Parameter	Type	Default	Description
aws_access_key_id	str	None	AWS Access Key ID (defaults to env var).
api_key	str	None	AWS Secret Access Key (defaults to env var).
region	str	None	AWS Region (e.g., "us-east-1").
sample_rate	int	16000	Sample rate (must be 8000 or 16000).
language	Language	EN	Language for transcription.

Usage

Basic Setup

import os
from piopiy.services.aws.stt import AWSTranscribeSTTService
from piopiy.transcriptions.language import Language

stt = AWSTranscribeSTTService(
    aws_access_key_id=os.getenv("AWS_ACCESS_KEY_ID"),
    api_key=os.getenv("AWS_SECRET_ACCESS_KEY"),
    region="us-east-1",
    language=Language.EN,
    sample_rate=16000
)

Notes

• Sample Rate: AWS Transcribe streaming only supports 8000 Hz or 16000 Hz. Piopiy will log a warning and default to 16000 Hz if an unsupported rate is provided.
• WebSocket Protocol: This service uses the HTTP/2 or WebSocket protocol for efficient, low-latency streaming.