DotNetTtsWrapper 1.1.2

DotNet TTS Wrapper

A .NET NuGet package that provides a unified API for working with multiple cloud-based and local Text-to-Speech (TTS) services. Ported from js-tts-wrapper.

Repository: https://github.com/AACTools/dotnet-tts-wrapper

Supported Engines

Features

• Unified API: Single interface for 20+ TTS engines
• True Streaming: IAsyncEnumerable-based audio chunk streaming where supported
• Word Timings: Real word boundary events from Azure SDK, estimated for other engines
• SSML Support: Fluent SSML builder for expressive speech synthesis
• Cross-platform: Windows, Linux, macOS support (engine-dependent)
• Modern .NET: Built for .NET 8.0 with latest C# language features

Installation

dotnet add package DotNetTtsWrapper

Basic Usage

using DotNetTtsWrapper.Models;
using DotNetTtsWrapper.Engines;

// Create a TTS client
var azureCredentials = new AzureCredentials 
{ 
    SubscriptionKey = "your-key", 
    Region = "westus" 
};
var client = TtsFactory.CreateClient("azure", azureCredentials);

// Get available voices
var voices = await client.GetVoicesAsync();
client.SetVoice(voices[0].Id);

// Synthesize speech
var result = await client.SynthToBytesAsync("Hello world!");
File.WriteAllBytes("output.wav", result.AudioData);

// With word timings
var options = new TtsOptions { EnableWordTimings = true };
var resultWithTimings = await client.SynthToBytesAsync("Hello world!", options);
foreach (var timing in resultWithTimings.WordTimings)
{
    Console.WriteLine($"{timing.Word}: {timing.StartTime}s - {timing.EndTime}s");
}

// Stream audio
var streamResult = await client.SynthToStreamAsync("Hello world!");
await foreach (var chunk in streamResult.AudioStream)
{
    // Process audio chunks
    ProcessAudioChunk(chunk.AudioData);
}

Advanced Features

SSML Builder

var ssml = SsmlBuilder.Speak()
    .Voice("en-US-AriaNeural")
    .WithRate("fast")
    .WithPitch("high")
    .WithVolume(80)
    .AddText("Hello world!")
    .Build();

Word Events

client.WordBoundary += (sender, e) => 
{
    Console.WriteLine($"Word: {e.Word}, Time: {e.StartTime}s");
};
await client.SpeakAsync("Hello world!");

Engine-Specific Setup

Azure Speech SDK

var credentials = new AzureCredentials 
{ 
    SubscriptionKey = "your-key", 
    Region = "your-region" 
};
var client = new AzureSdkTtsClient(credentials);

SAPI (Windows Only)

var client = new SapiTtsClient();
var voices = await client.GetVoicesAsync();

Architecture

• Abstract Factory Pattern: TtsFactory.CreateClient() for engine creation
• Event-Driven: Word boundary, speech started/completed events
• Async-First: Full async/await support throughout
• Streaming: IAsyncEnumerable for true audio chunk streaming
• Modular: Optional engine-specific packages for reduced dependencies

Requirements

• .NET 8.0 or higher
• Windows OS for SAPI engine
• Platform-specific packages for some engines

License

Ported from js-tts-wrapper with .NET-specific enhancements.

Roadmap

• eSpeak integration (local offline TTS)
• Speech Markdown support
• Advanced model management for SherpaOnnx
• Additional cloud engine integrations