sonusai-asr-cloud 1.3.0

Cloud ASR tools for SonusAI

SonusAI ASR Cloud

Cloud ASR tools for SonusAI.

SonusAI ASR Cloud provides provider-specific adapters for cloud speech-to-text services and normalizes responses into SonusAI ASR result types.

Key Features

• Multiple Cloud Providers: aiXplain Whisper, Deepgram, and SpeechRecognition (Google Cloud Speech API).
• Unified Return Type: Provider outputs are normalized to ASRResult.
• Validation Helpers: Each provider includes config/environment validation helpers.
• SonusAI Integration: Uses SonusAI datatypes and utility helpers for audio conversion and output formatting.

Getting Started

Prerequisites

• Python 3.13 or 3.14
• uv

Install

uv sync
source .venv/bin/activate

Provider Requirements

• aiXplain Whisper
  • Pass model in provider config.
  • Set environment variable: AIXP_WHISPER_<MODEL>.
• Deepgram
  • Set environment variable: DEEPGRAM_API_KEY.
• Google (SpeechRecognition)
  • Configure runtime credentials for Google Cloud Speech as required by the underlying backend.

Usage Notes

Provider functions live under sonusai_asr_cloud/asr_functions/ and expose:

• *_validate(**config) for preflight checks.
• provider call function (for example deepgram(audio, **config)) returning ASRResult.

Development

• Development Guide: docs/development.md
• Agent/Contributor Guidance: AGENTS.md

Common Commands

pytest
./build_wheels.sh
./publish.sh

Project Context

• Overview
• Architecture
• Standards
• Directory Map
• CLI & Commands