A cross-platform Rust engine for neural TTS models.

• Piper

• espeak-phonemizer: Converts text to IPA phonemes using a patched version of eSpeak-ng
• sonata-model: Handles model loading and inference using onnxruntime via ort
• sonata-synth: Wraps SonataModel and adds synthesized speech post-processing, including changing prosody. Also provides different modes of parallelism.
• sonata-grpc: GRPC frontend for sonata
• libsonata: C-API binding to sonata
• sonata-python: Python bindings to sonata-synth using pyo3
• sonic-sys: Rust FFI bindings to Sonic: a C library for controlling various aspects of generated speech, such as rate, volume, and pitch

Some packages, such as espeak-phonemizer, include tests. Running cargo test from the root of the workspace will likely fail, because cargo does not load config from sub packages when ran from the workspace root.

On Windows you need to add espeak-ng.dll to the library search path by modifying the PATH environment variable.

For example, to add espeak-ng.dll to your path when building for the x86_64-pc-windows-msvc target, run the following command before cargo test:

set PATH=%PATH%;{repo_path}\deps\windows\espeak-ng-build\i686\bin

Replace repo_path with the absolute path to the repository.

Then cd to the package, and run cargo test from there.

Copyright (c) 2023 Musharraf Omer. This code is licensed under the MIT license.