SpeechMarkdown Rust

High-performance SpeechMarkdown parser written in Rust. Converts SpeechMarkdown syntax to platform-specific SSML for Amazon Alexa, Google Assistant, Microsoft Azure, and more.

Repository Structure

This repository contains the core Rust SpeechMarkdown parser with bindings for multiple languages:

speechmarkdown-rust/
├── src/                      # Core Rust library (language-agnostic)
├── bindings/                 # Language-specific bindings
│   ├── python/               # Python bindings via PyO3
│   ├── nodejs/               # Node.js bindings via napi-rs
│   ├── dotnet/               # .NET bindings via rust-cpp
│   └── swift/                # Swift bindings via C API + SPM
├── Package.swift             # Swift Package Manager configuration (Swift-only)
├── build-swift-package.sh    # Swift package build script (Swift-only)
└── Cargo.toml                # Rust package configuration

Important Notes for Non-Swift Users:

• Swift-specific files are ignored by other build systems: Package.swift, build-swift-package.sh, and swift-package-dist/ are only used by Swift Package Manager. They do not affect Rust, Python, Node.js, or .NET builds.
• Core Rust library unchanged: The src/ directory and Cargo.toml remain the single source of truth for the SpeechMarkdown parser across all languages.
• Isolated language bindings: Each language in bindings/ has its own build configuration and doesn't interfere with others.

For Rust/.NET/Python/Node.js Developers: You can safely ignore Swift-specific files. Your respective package managers (Cargo, NuGet, PyPI, npm) only use the core library and your language's binding directory.

Install

Supported Platforms

API

All bindings expose the same core methods:

Usage

Rust

use speechmarkdown_rust::{SpeechMarkdownParser, Platform};

// Convert to SSML
let ssml = SpeechMarkdownParser::to_ssml(
    "Hello (world)[emphasis:\"strong\"]",
    Platform::AmazonAlexa,
)?;

// Convert to plain text
let text = SpeechMarkdownParser::to_text("Hello (world)[emphasis:\"strong\"]")?;

// Parse to AST (JSON string)
let ast = SpeechMarkdownParser::parse("Hello world")?;

// Check if input contains SpeechMarkdown syntax
if SpeechMarkdownParser::is_speech_markdown(&input) {
    // ...
}

// Validate input
SpeechMarkdownParser::validate(&input)?;

// Convert SSML back to SpeechMarkdown (best-effort)
let smd = SpeechMarkdownParser::to_smd(r#"<speak><emphasis level="strong">word</emphasis></speak>"#)?;
// Returns: ++word++

Python

from speechmarkdown_rust import to_ssml, to_text, parse, is_speech_markdown, validate

ssml = to_ssml('Hello (world)[emphasis:"strong"]', 'amazon-alexa')
text = to_text('Hello (world)[emphasis:"strong"]')
ast = parse('Hello world')

is_smd = is_speech_markdown('Hello (world)[emphasis:"strong"]')  # True
is_smd = is_speech_markdown('Hello world')                       # False

validate('Hello (world)[emphasis:"strong"]')  # raises ValueError if invalid

# Convert SSML to SpeechMarkdown (best-effort)
smd = to_smd('<speak><emphasis level="strong">word</emphasis></speak>')
# Returns: ++word++

Node.js

const { to_ssml, to_text, parse, is_speech_markdown, validate } = require('speechmarkdown')

const ssml = to_ssml('Hello (world)[emphasis:"strong"]', 'amazon-alexa')
const text = to_text('Hello (world)[emphasis:"strong"]')
const ast = parse('Hello world')

is_speech_markdown('Hello (world)[emphasis:"strong"]')  // true
is_speech_markdown('Hello world')                       // false

validate('Hello (world)[emphasis:"strong"]')  // throws if invalid

// Convert SSML to SpeechMarkdown (best-effort)
const smd = to_smd('<speak><emphasis level="strong">word</emphasis></speak>')
// Returns: ++word++

.NET (C#)

using SpeechMarkdown;

var parser = new SpeechMarkdownParser();

string ssml = parser.ToSsml("Hello (world)[emphasis:\"strong\"]", Platform.AmazonAlexa);
string text = parser.ToText("Hello (world)[emphasis:\"strong\"]");
string json = parser.ParseToJson("Hello world");

bool isSmd = parser.IsSpeechMarkdown("Hello (world)[emphasis:\"strong\"]"); // true
parser.Validate("Hello (world)[emphasis:\"strong\"]"); // throws on invalid

// Convert SSML to SpeechMarkdown (best-effort)
string smd = parser.ToSmd("<speak><emphasis level=\"strong\">word</emphasis></speak>");
// Returns: ++word++

Swift

The repo root contains a Package.swift with a binary target pointing to a pre-built XCFramework. This enables both local development and Swift Package Index integration.

Option 1 — SPM (recommended):

// In your Package.swift:
.package(url: "https://github.com/AACTools/speechmarkdown-rust", branch: "spm")

Option 2 — Local package: Download speechmarkdown-swift-package.zip from the latest release, unzip, and add as a local package in Xcode (File > Add Packages > Add Local).

Includes macOS (arm64 + x86_64), iOS device (arm64), and iOS simulator (arm64) slices.

To build from source: ./build-swift-package.sh

import SpeechMarkdown

let parser = SpeechMarkdownParser()

let ssml = try parser.toSsml(input: "Hello (world)[emphasis:\"strong\"]", platform: "amazon-alexa")
let text = try parser.toText(input: "Hello (world)[emphasis:\"strong\"]")
let json = try parser.parseToJson(input: "Hello world")

let isSmd = parser.isSpeechMarkdown(input: "Hello (world)[emphasis:\"strong\"]") // true
try parser.validate(input: "Hello (world)[emphasis:\"strong\"]") // throws on invalid

// Convert SSML to SpeechMarkdown (best-effort)
let smd = try parser.toSmd(ssml: "<speak><emphasis level=\"strong\">word</emphasis></speak>")
// Returns: ++word++

C API

#include "speechmarkdown.h"

// Convert to SSML
const char* ssml = speechmarkdown_to_ssml("Hello (world)[emphasis:\"strong\"]", "amazon-alexa");
speechmarkdown_free((char*)ssml);

// Convert to plain text
const char* text = speechmarkdown_to_text("Hello (world)[emphasis:\"strong\"]");
speechmarkdown_free((char*)text);

// Parse to JSON
const char* json = speechmarkdown_parse("Hello world");
speechmarkdown_free((char*)json);

// Check for SpeechMarkdown syntax
bool is_smd = speechmarkdown_is_speech_markdown("Hello (world)[emphasis:\"strong\"]");

// Validate
bool valid = speechmarkdown_validate("Hello (world)[emphasis:\"strong\"]");

// Convert SSML to SpeechMarkdown (best-effort)
const char* smd = speechmarkdown_to_smd("<speak><emphasis level=\"strong\">word</emphasis></speak>");
speechmarkdown_free((char*)smd);

// Get last error (thread-local)
const char* err = speechmarkdown_get_error();

Building from Source

Core Rust Library

The core Rust parser is built with:

cargo build --release

This produces:

• Windows: target/release/speechmarkdown_rust.dll
• macOS: target/release/libspeechmarkdown_rust.dylib
• Linux: target/release/libspeechmarkdown_rust.so

Language-Specific Builds

Each language binding has its own build process:

Python:

cd bindings/python
maturin develop
# or for release: maturin build --release

Node.js:

cd bindings/nodejs
npm install
npm run build

.NET:

cd bindings/dotnet
dotnet build

Swift:

# Uses pre-built XCFramework from releases
# Or build from source:
./build-swift-package.sh

Development Workflow

The repository follows a unified development model where all language bindings share the same core Rust parser:

1. Core changes: Modify src/ for parser logic changes
2. Language changes: Modify bindings/<language>/ for binding-specific changes
3. Testing: Each language has its own test suite
4. Releases: All packages are released together with version synchronization

For Contributors:

• Rust developers: Work in src/ and test with cargo test
• Python developers: Work in bindings/python/ and test with pytest
• Node.js developers: Work in bindings/nodejs/ and test with npm test
• .NET developers: Work in bindings/dotnet/ and test with dotnet test
• Swift developers: Work in bindings/swift/ and use SPM for testing

License

MIT