opea/erag-asr repository overview

⁠OPEA ERAG ASR Microservice

Part of the Intel® AI for Enterprise RAG (ERAG) ecosystem.

⁠🔍 Overview

The OPEA ERAG ASR (Automatic Speech Recognition) microservice converts audio files to text using Whisper or compatible ASR models. It interfaces with ASR model servers to process audio and return transcriptions.

⁠Support Matrix

⁠Features

• Converts audio files to text using Whisper ASR models
• Supports multiple audio formats (WAV, MP3)
• OpenAI-compatible API interface
• Configurable language detection and model selection

⁠🔗 Related Components

This service integrates with other OPEA ERAG components:

• OPEA ERAG Text Splitter can divide the transcribed text into chunks for data ingestion
• OPEA ERAG Retriever & Reranker microservices can use transcriptions for document retrieval
• It triggers transcription requests with vLLM Model Server running on Intel® Xeon® Processors

⁠License

OPEA ERAG is licensed under the Apache License, Version 2.0.

Copyright © 2024–2026 Intel Corporation. All rights reserved.