speech-swift

Overview

Speech Swift from Soniqo is a set of AI speech models for Apple silicon, built on MLX Swift and Core ML. Each model is its own Swift package library, so you only include what you need. It covers speech to text, text to speech, translation, voice activity detection, speaker diarisation and more.

Features

• Speech to text, including streaming, and forced alignment
• Text to speech and speech to speech
• Text and speech translation
• Voice activity detection and speaker diarisation
• Speech enhancement and acoustic echo cancellation
• Optional SwiftUI views and demo apps

Requirements

• Swift 6 or later and Xcode 16 or later with the Metal Toolchain
• macOS 15 (Sequoia) or later, or iOS 18 or later
• Apple silicon (M1 or later)
• Native ARM Homebrew for the Homebrew install

Install

1. Add the package to Package.swift and choose the model products you need

   .package(url: "https://github.com/soniqo/speech-swift", branch: "main")

2. Or install with Homebrew

   brew install speech