Expand description
Product OS : Voice — text-to-speech and voice generation with pluggable backends.
This library provides a unified API for synthesizing speech from text. Multiple backends can be enabled via feature flags: local TTS (Piper, eSpeak NG, Edge TTS) or cloud (AWS Polly).
Library-handled setup: The piper (subprocess) backend is fully handled by the library—no
shell scripts. Call create_engine with "piper" and None config to
trigger download and setup. This works without piper-rs. The piper_embedded backend is
optional and may have dependency constraints (e.g. piper-rs, cmake for first build).
§Features
- piper: Piper TTS (subprocess; binary and ONNX model auto-downloaded when config is
None). Recommended. - piper_embedded: Piper TTS in-process via piper-rs (optional; dependency compatibility may vary).
- kokoro_embedded: Kokoro TTS in-process via ONNX Runtime + eSpeak phonemization (optional).
- sherpa_embedded: Sherpa-ONNX offline VITS via C API + same ONNX Runtime as Piper/Kokoro (optional; requires native Sherpa libs).
- espeak: eSpeak NG (requires
espeak-ngbinary on PATH). - edge_tts: Edge TTS (requires
edge-ttson PATH orpython3withpip install edge-tts; no API key). - aws_polly: AWS Polly (requires AWS credentials and network).
No backend is enabled by default; add the feature you need.
§Example
use product_os_voice::{create_engine, VoiceEngine, VoiceOptions};
use std::path::PathBuf;
let engine = product_os_voice::create_engine("piper", None).expect("piper feature enabled");
// Piper requires a model; in real use configure with PiperEngine::new().with_model(path)
let options = VoiceOptions::default();
let audio = engine.synthesize("Hello, world!", &options);
if let Ok(bytes) = audio {
std::fs::write("output.wav", bytes).expect("write");
}§Doc-test
Building options and error types (no backend required):
use product_os_voice::{AudioFormat, VoiceOptions};
let opts = VoiceOptions {
sample_rate: Some(16000),
format: AudioFormat::Wav,
..Default::default()
};
assert_eq!(opts.sample_rate, Some(16000));Re-exports§
pub use config::EngineConfig;pub use engine::AudioFormat;pub use engine::VoiceEngine;pub use engine::VoiceInfo;pub use engine::VoiceOptions;pub use error::VoiceError;
Modules§
- backends
- Voice engine backends (local and cloud).
- config
- Engine configuration for backends that require credentials or paths.
- engine
- Core types and trait for voice synthesis.
- error
- Error types for voice synthesis operations.
- platform
- Platform-specific behavior for voice backends (macOS, iOS, Android).
Functions§
- create_
engine - Create a voice engine by name, with optional backend-specific config.