Skip to main content

Azure AI Speech (Cognitive Services)

Azure AI Speech is Azure's Cognitive Services text-to-speech API, separate from Azure OpenAI. It provides high-quality neural voices with broader language support and advanced speech customization.

When to use this vs Azure OpenAI TTS:

  • Azure AI Speech - More languages, neural voices, SSML support, speech customization
  • Azure OpenAI TTS - OpenAI models, integrated with Azure OpenAI services

Overview​

PropertyDetails
DescriptionAzure AI Speech is Azure's Cognitive Services text-to-speech API, separate from Azure OpenAI. It provides high-quality neural voices with broader language support and advanced speech customization.
Provider Route on LiteLLMazure/speech/

Quick Start​

LiteLLM SDK

SDK Usage
from litellm import speech
from pathlib import Path
import os

os.environ["AZURE_TTS_API_KEY"] = "your-cognitive-services-key"

speech_file_path = Path(__file__).parent / "speech.mp3"
response = speech(
    model="azure/speech/azure-tts",
    voice="alloy",
    input="Hello, this is Azure AI Speech",
    api_base="https://eastus.tts.speech.microsoft.com",
    api_key=os.environ["AZURE_TTS_API_KEY"],
)
response.stream_to_file(speech_file_path)

LiteLLM Proxy

proxy_config.yaml
model_list:
  - model_name: azure-speech
    litellm_params:
      model: azure/speech/azure-tts
      api_base: https://eastus.tts.speech.microsoft.com
      api_key: os.environ/AZURE_TTS_API_KEY

Setup​

  1. Create an Azure Cognitive Services resource in the Azure Portal
  2. Get your API key from the resource
  3. Note your region (e.g., eastus, westus, westeurope)
  4. Use the regional endpoint: https://{region}.tts.speech.microsoft.com

Voice Mapping​

LiteLLM automatically maps OpenAI voice names to Azure Neural voices:

OpenAI VoiceAzure Neural VoiceDescription
alloyen-US-JennyNeuralNeutral and balanced
echoen-US-GuyNeuralWarm and upbeat
fableen-GB-RyanNeuralExpressive and dramatic
onyxen-US-DavisNeuralDeep and authoritative
novaen-US-AmberNeuralFriendly and conversational
shimmeren-US-AriaNeuralBright and cheerful

Supported Parameters​

All Parameters
response = speech(
    model="azure/speech/azure-tts",
    voice="alloy",                    # Required: Voice selection
    input="text to convert",          # Required: Input text
    speed=1.0,                        # Optional: 0.25 to 4.0 (default: 1.0)
    response_format="mp3",            # Optional: mp3, opus, wav, pcm
    api_base="https://eastus.tts.speech.microsoft.com",
    api_key="your-key",
)

Response Formats​

FormatAzure Output FormatSample Rate
mp3audio-24khz-48kbitrate-mono-mp324kHz
opusogg-48khz-16bit-mono-opus48kHz
wavriff-24khz-16bit-mono-pcm24kHz
pcmraw-24khz-16bit-mono-pcm24kHz

Async Support​

Async Usage
import asyncio
from litellm import aspeech
from pathlib import Path

async def generate_speech():
    response = await aspeech(
        model="azure/speech/azure-tts",
        voice="alloy",
        input="Hello from async",
        api_base="https://eastus.tts.speech.microsoft.com",
        api_key=os.environ["AZURE_TTS_API_KEY"],
    )
    
    speech_file_path = Path(__file__).parent / "speech.mp3"
    response.stream_to_file(speech_file_path)

asyncio.run(generate_speech())

Regional Endpoints​

Replace {region} with your Azure resource region:

  • US East: https://eastus.tts.speech.microsoft.com
  • US West: https://westus.tts.speech.microsoft.com
  • Europe West: https://westeurope.tts.speech.microsoft.com
  • Asia Southeast: https://southeastasia.tts.speech.microsoft.com

Full list of regions

Advanced Features​

Custom Neural Voices​

You can use any Azure Neural voice by passing the full voice name:

Custom Voice
response = speech(
    model="azure/speech/azure-tts",
    voice="en-US-AriaNeural",  # Direct Azure voice name
    input="Using a specific neural voice",
    api_base="https://eastus.tts.speech.microsoft.com",
    api_key=os.environ["AZURE_TTS_API_KEY"],
)

Browse available voices in the Azure Speech Gallery.

Error Handling​

Error Handling
from litellm import speech
from litellm.exceptions import APIError

try:
    response = speech(
        model="azure/speech/azure-tts",
        voice="alloy",
        input="Test message",
        api_base="https://eastus.tts.speech.microsoft.com",
        api_key=os.environ["AZURE_TTS_API_KEY"],
    )
except APIError as e:
    print(f"Azure Speech error: {e}")

Reference​