Transcribe Audio with Azure AI

Azure AI Transcription SDK for Python provides a client library for converting speech to text using Azure Cognitive Services. It supports both batch transcription for audio files stored in blob storage and real-time streaming transcription, with optional speaker diarization to identify multiple speakers in recordings.

Use this skill when you need to transcribe meeting recordings with multiple speakers, generate subtitles from audio files with timestamps, or process real-time audio streams into text. Enable diarization when multiple speakers are present, use batch transcription for long files stored in blob storage, and specify language locale to improve recognition accuracy.

Do not use this skill when the task falls outside speech-to-text transcription workflows. Do not proceed if required Azure Cognitive Services credentials (endpoint and subscription key) are missing, as DefaultAzureCredential is not supported for this client.

You provide an Azure Cognitive Services endpoint URL, a subscription key, audio content (either as URLs pointing to blob storage for batch jobs or audio file streams for real-time processing), and a locale identifier (e.g., "en-US"). For batch transcription, you can optionally enable diarization and provide a job name.

You receive transcription results with status information for batch jobs, or streaming text events for real-time transcription.

This skill uses Azure Cognitive Services and accepts audio file URLs for batch transcription workflows.

Install the library:

pip install azure-ai-transcription

Configure environment variables:

TRANSCRIPTION_ENDPOINT=https://<resource>.cognitiveservices.azure.com
TRANSCRIPTION_KEY=<your-key>

Authenticate using subscription key:

import os
from azure.ai.transcription import TranscriptionClient

client = TranscriptionClient(
    endpoint=os.environ["TRANSCRIPTION_ENDPOINT"],
    credential=os.environ["TRANSCRIPTION_KEY"]
)

For batch transcription:

job = client.begin_transcription(
    name="meeting-transcription",
    locale="en-US",
    content_urls=["https://<storage>/audio.wav"],
    diarization_enabled=True
)
result = job.result()
print(result.status)

For real-time streaming:

stream = client.begin_stream_transcription(locale="en-US")
stream.send_audio_file("audio.wav")
for event in stream:
    print(event.text)

This skill is for users who need to transcribe speech to text using Azure Cognitive Services with Python.