Skip to main content

Microsoft Azure TTS

Microsoft Azure Cognitive Services Text-to-Speech provides high-quality voice synthesis with support for neural voices, custom voice creation, and extensive SSML features.

Authentication​

Azure TTS requires a subscription key and region:

from tts_wrapper import MicrosoftTTS, MicrosoftClient

client = MicrosoftClient(credentials=(
    'subscription_key',  # Your Azure subscription key
    'region'            # e.g., 'eastus', 'westeurope'
))

tts = MicrosoftTTS(client)
tip

Use environment variables for secure credential management:

import os

client = MicrosoftClient(credentials=(
    os.getenv('MICROSOFT_TOKEN'),
    os.getenv('MICROSOFT_REGION')
))

Features​

SSML Support​

Azure TTS provides extensive SSML support with unique features:

ssml_text = """
<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis"
       xmlns:mstts="http://www.w3.org/2001/mstts">
    Hello <break time="300ms"/> World!
    <prosody rate="-20%" pitch="+20%">
        This is a test of SSML features.
    </prosody>
    <mstts:express-as style="cheerful">
        This text will be spoken in a cheerful way!
    </mstts:express-as>
</speak>
"""
tts.speak(ssml_text)

See Microsoft TTS SSML Reference for all supported tags.

Streaming​

Supports real-time audio streaming:

# Stream synthesis for real-time playback
tts.speak_streamed("This text will be synthesized and played in real-time")

Word Timing​

Get precise timing information for each word:

def on_word(word: str):
    print(f"Speaking: {word}")

tts.connect("started-word", on_word)
tts.speak("This text will trigger word timing callbacks")

Voice Selection​

List and select from available voices:

# Get list of available voices
voices = tts.get_voices()
for voice in voices:
    print(f"Name: {voice['name']}")
    print(f"Locale: {voice['language_codes'][0]}")
    print(f"Gender: {voice['gender']}")
    print("---")

# Set a specific voice
tts.set_voice("en-US-JennyNeural", "en-US")

Voice Styles and Roles​

Azure offers unique voice styling capabilities:

ssml_text = """
<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis"
       xmlns:mstts="http://www.w3.org/2001/mstts">
    <mstts:express-as style="excited" styledegree="2">
        This is very exciting news!
    </mstts:express-as>
</speak>
"""
tts.speak(ssml_text)

Best Practices​

  1. Cost Management

    • Use streaming for long text
    • Monitor usage through Azure Portal
    • Cache frequently used phrases

  2. Performance

    • Reuse client instances
    • Choose appropriate region
    • Use appropriate voice selection for your needs

  3. Error Handling

    try:
        tts.speak("Hello, world!")
    except Exception as e:
        if "AuthenticationFailed" in str(e):
            print("Check your Azure credentials")
        elif "QuotaExceeded" in str(e):
            print("Usage quota exceeded")
        else:
            print(f"Error: {e}")

Limitations​

  • Maximum text length varies by endpoint
  • API rate limits apply (check Azure quotas)
  • Some voice styles only available with specific voices
  • Neural voices may have higher latency
  • Custom voice creation requires additional setup

Additional Resources​

Next Steps​