FastSSML 0.1.0 [ALPHA-2026-10] — High-Performance, Zero-Overhead SSML Dialect Engine for Java

---

⚡ High-Performance, Zero-Allocation SSML Builder & Cross-Engine Speech Dialect Engine for Java.

FastSSML provides ultra-fast speech markup construction and multi-engine dialect translation directly on the JVM with zero external runtime bloat. It synthesizes standards-compliant W3C SSML, Microsoft Edge TTS markup, Windows SAPI/OneCore speech strings, Deepgram Aura formats, and raw plain-text speech strips with deterministic sub-microsecond throughput.

---

Quick Start — Example

import fastssml.FastSSML;
import fastssml.FastSSML.Emphasis;
import fastssml.FastSSML.Dialect;

public class Demo {
    public static void main(String[] args) {
        // 1. Build expressive, entity-safe speech markup
        FastSSML speech = FastSSML.create()
            .voice("de-DE-FlorianMultilingualNeural")
            .lang("de-DE")
            .rate(1.10f)     // +10% speed
            .pitch(1.05f)    // +5% pitch
            .volume(0.95f)   // -5% volume
            .text("Willkommen beim FastJava Ecosystem! ")
            .emphasis("High-Performance", Emphasis.STRONG)
            .pause(250)      // 250ms break
            .text(" 0 Kopien & maximale Sprachqualität.");

        // 2. Render to standard W3C SSML or target dialect
        String ssml = speech.toSSML();
        System.out.println(ssml);

        // 3. Or strip tags for plain-text TTS engines (e.g. ElevenLabs / Piper)
        String plain = speech.toPlainText();
        System.out.println(plain);
    }
}

---

Table of Contents

• Why FastSSML?
• Key Features
• Architecture & Dialect Pipeline
• Performance & Benchmarks
• Multi-Engine Dialect Matrix
• Installation
• Technical Demos & Benchmarks
• Documentation
• Platform Support
• License
• Related Projects

---

Why FastSSML?

Traditional Java approaches to Speech Synthesis Markup Language (SSML) construction suffer from severe performance bottlenecks and runtime fragility:

1. Heavyweight XML DOM Trees: Libraries relying on javax.xml.parsers.DocumentBuilder or JAXB allocate hundreds of heap objects per sentence, triggering GC churn and latency spikes.
2. String Concatenation Vulnerabilities: Manually assembling SSML via String.format or string templates risks unescaped XML entity injection (&, <, >, ", '), breaking synthesis engines mid-stream.
3. Fragmented Dialect Formats: Speech engines do not share a single SSML standard. Edge TTS requires <prosody>, Windows SAPI uses custom rate percentages, while ElevenLabs or local ONNX Piper models require clean stripped plain text.
4. Unnecessary Native Runtime Bloat: Markup builders shouldn't require JNI bindings or native binaries. FastSSML runs as 100% pure Java 17+ with zero external dependencies.

Feature	Legacy XML Builders (DOM/JAXB)	Manual String Templates	FastSSML
Memory Allocation	High (~20–80 KB / doc)	Medium (Excessive Strings)	Minimal (Pre-sized Buffer)
Throughput	~5,000 ops/sec	~40,000 ops/sec	> 200,000–1,000,000 ops/sec
XML Entity Safety	✅ Safe	❌ Vulnerable to injection	✅ Instant Entity Escaping
Multi-Engine Dialects	❌ Standard only	❌ Hardcoded per engine	✅ Built-in Dialect Engine
External Dependencies	Heavy (xerces, jaxb)	None	Zero Bloat (Pure Java 17+)
GC Pressure	Extreme	Moderate	Near-Zero Allocation

---

Key Features

• 🗣️ Cross-Engine Speech Dialects — One fluent builder generates W3C Standard SSML, Edge TTS, Windows SAPI, Deepgram, and ElevenLabs plain-text formats on demand.
• ⚡ Sub-Microsecond Generation — Optimized character streaming directly writes XML tags without tree traversal or reflection.
• 🛡️ Automated XML Entity Escaping — Instant sanitization of ampersands, angle brackets, and quotes prevents broken synthesis payloads.
• 🎛️ Complete Prosody Control — Fine-grained management over voice, lang, rate (speed), pitch, volume, pause (<break>), and emphasis.
• 🪶 Zero Native Footprint — 100% pure Java 17+. No DLLs, no JNI, no compile step. Works out-of-the-box on Windows, Linux, and macOS.
• 🎯 FastTTS & FastAudio Synergy — Pairs directly with FastTTS and FastAudioPlayer for unified, low-latency audio pipelines.

---

Architecture & Dialect Pipeline

┌─────────────────────────────────────────────────────────────┐
│                    Java Application Code                    │
└──────────────────────────────┬──────────────────────────────┘
                               │ Fluent API Calls
                               ▼
┌─────────────────────────────────────────────────────────────┐
│               FastSSML Fluent Builder Engine                │
│    (Character Stream Escaping & Pre-sized Buffering)        │
└──────────────────────────────┬──────────────────────────────┘
                               │ toSSML(Dialect)
        ┌──────────────────────┼──────────────────────┐
        ▼                      ▼                      ▼
┌──────────────┐       ┌──────────────┐       ┌──────────────┐
│   Standard   │       │   Edge TTS   │       │ Plain Text   │
│   W3C SSML   │       │   & SAPI     │       │ (Stripped)   │
└──────────────┘       └──────────────┘       └──────────────┘
        │                      │                      │
        ▼                      ▼                      ▼
┌─────────────────────────────────────────────────────────────┐
│               FastTTS / Speech Synthesis Engine             │
│        (Windows SAPI, Piper ONNX, Edge, ElevenLabs)         │
└─────────────────────────────────────────────────────────────┘

---

Performance & Benchmarks

Measured on OpenJDK 21 LTS, Windows 11 x64:

Operation	Throughput	Latency	Allocation Churn
Standard SSML Build	~200,000+ docs/sec	< 5 µs	Pre-sized buffer
Dialect Translation (Edge TTS)	~250,000+ docs/sec	< 4 µs	Single-pass assembly
Plain-Text Extraction (toPlainText)	~500,000+ ops/sec	< 2 µs	Minimal GC

---

Multi-Engine Dialect Matrix

Engine	Dialect Target	Supported Prosody	Tag Format
W3C Standard	Dialect.STANDARD	Rate, Pitch, Volume, Break, Emphasis	Standard <speak> XML
Microsoft Edge TTS	Dialect.EDGE_TTS	Neural Voice, Prosody Percentage	Microsoft Speech Synthesizer XML
Windows SAPI / OneCore	Dialect.WINDOWS_SAPI	System Voice, Standard Prosody	Native Win32 SAPI XML
Deepgram Aura	Dialect.DEEPGRAM	Standard Speech Markup	Aura Voice Endpoints
ElevenLabs / Piper	Dialect.ELEVENLABS_PLAIN	Plain text output	Automatic tag-stripping

---

Installation

Option 1: Maven (Recommended via JitPack)

Add the JitPack repository and dependency to your pom.xml:

<repositories>
    <repository>
        <id>jitpack.io</id>
        <url>https://jitpack.io</url>
    </repository>
</repositories>

<dependencies>
    <dependency>
        <groupId>com.github.andrestubbe</groupId>
        <artifactId>FastSSML</artifactId>
        <version>0.1.0</version>
    </dependency>
</dependencies>

Option 2: Gradle (via JitPack)

Add this to your build.gradle:

repositories {
    maven { url 'https://jitpack.io' }
}

dependencies {
    implementation 'com.github.andrestubbe:FastSSML:0.1.0'
}

Option 3: Direct Download (Pre-built JAR)

Download the latest pre-compiled JAR directly from GitHub Releases:

1. 📦 FastSSML-0.1.0.jar

---

Technical Demos & Benchmarks

Case	Java Example	Launcher	Description
Interactive Showcase Demo	Demo.java	run-demo.bat	End-to-end demonstration of SSML generation, XML escaping, dialect rendering, and micro-throughput test.
JMH Microbenchmark Suite	Benchmark.java	run-benchmark.bat	Formal OpenJDK JMH throughput measurements across standard SSML, Edge TTS, and plain-text stripping.

---

Documentation

• REFERENCE.md: Exhaustive catalog of API methods, builder contracts, and dialect specifications.
• PHILOSOPHY.md: Architectural principles: zero-allocation builder, pure Java design, and dialect normalization.
• ROADMAP.md: Planned milestone features and SSML 1.1 extension support.
• CHANGELOG.md: Release history and version documentation.
• COMPILE.md: Compilation guide and test execution.

---

Platform Support

FastSSML is a pure Java library with zero native binary requirements. It runs identically across all platforms:

Platform	Architecture	Status	Notes
Windows 10/11	x64, ARM64	✅ Fully Supported	100% Pure Java 17+
Linux	x64, ARM64	✅ Fully Supported	100% Pure Java 17+
macOS	Apple Silicon, x64	✅ Fully Supported	100% Pure Java 17+

---

License

MIT License — See LICENSE for details.

---

Related Projects

• FastCore — Native library loader, FFM gateway, and platform abstraction
• FastTTS — Unified, zero-bloat text-to-speech backend orchestration
• FastSTT — High-throughput speech-to-text integration for Java
• FastAudioCapture — Low-latency WASAPI loopback and microphone audio capture
• FastAudioPlayer — Native low-latency audio playback for Java via WASAPI
• FastAudioProcess — Zero-allocation audio DSP, resamplers, and format converters
• FastVAD — Native Voice Activity Detection (Silero ONNX / WebRTC)
• FastTurn — Real-time conversational turn-taking and speech arbitration
• FastWakeWord — Zero-latency offline wake word detection engine

---

Part of the FastJava Ecosystem — Making the JVM faster. Small package. Maximum speed. Zero bloat. 🚀📋