Azure AI Speech

Downloads

Frameworks

.NET

.NET MAUI

Operating Systems

Android

iOS

Windows

Overview

Azure AI Speech provides enterprise-grade cloud speech-to-text and text-to-speech powered by Azure Cognitive Services. It replaces the platform-native ISpeechToTextService and ITextToSpeechService registrations with cloud-backed implementations while still using platform-native audio capture and playback.

Setup

• NuGet Packages
• Project File
• MauiProgram.cs
• Package.appxmanifest

Shiny.Speech.Azure

// In MauiProgram.cs
builder.Services.AddAzureSpeech("your-subscription-key", "eastus");
// IAudioSource and IAudioPlayer are automatically registered for platform audio I/O

Or with a config object and selective services:

builder.Services.AddAzureSpeech(
    new AzureSpeechConfig
    {
        SubscriptionKey = "your-subscription-key",
        Region = "eastus"
    },
    speechToText: true,
    textToSpeech: true
);

Configuration

public record AzureSpeechConfig
{
    public required string SubscriptionKey { get; init; }
    public required string Region { get; init; }
}

Property	Description
SubscriptionKey	Your Azure Speech Services subscription key
Region	Azure region (e.g., eastus, westus2, westeurope)

Usage

Once registered, inject and use ISpeechToTextService and ITextToSpeechService exactly as you would with platform-native speech — the API is identical:

public class MyViewModel(ISpeechToTextService stt, ITextToSpeechService tts)
{
    async Task ListenAndRespond(CancellationToken ct)
    {
        var access = await stt.RequestAccess();
        if (access != AccessState.Available)
            return;

        var text = await stt.ListenUntilSilence(cancellationToken: ct);

        if (text != null)
            await tts.SpeakAsync($"You said: {text}");
    }
}

Emotion & Tone

Azure expresses emotion through SSML, so a portable SpeechTone becomes an mstts:express-as wrapper and any bracketed annotations in the text are stripped rather than spoken aloud:

await tts.SpeakAsync("We just shipped it.", new TextToSpeechOptions
{
    Tone = new SpeechTone { Emotion = SpeechEmotion.Excited, Intensity = 1.5f }
});

<prosody rate="+0%" pitch="+0%" volume="100">
    <mstts:express-as style="excited" styledegree="1.5">
        We just shipped it.
    </mstts:express-as>
</prosody>

Intensity maps to styledegree (clamped to Azure’s 0.01–2 range). Style support is per-voice — a voice that doesn’t offer the requested style renders in its default delivery rather than failing, so check the voice list before relying on a particular style. No wrapper is emitted at all when there’s no tone, leaving the SSML byte-identical to before.

STT-Only or TTS-Only

You can register Azure for just one service:

// Azure STT only (use platform-native TTS)
builder.Services.AddTextToSpeech();   // Platform-native TTS
builder.Services.AddAzureSpeech("key", "region", speechToText: true, textToSpeech: false);

// Azure TTS only (use platform-native STT)
builder.Services.AddSpeechToText();   // Platform-native STT
builder.Services.AddAzureSpeech("key", "region", speechToText: false, textToSpeech: true);

Links

• Azure Speech Services Documentation
• Azure Speech Pricing