sherpa-onnx-tts

The skill downloads the sherpa-onnx runtime binary for your platform (macOS, Linux x64, or Windows x64) and a VITS voice model to local directories.A wrapper script exposes a simple CLI interface: sherpa-onnx-tts -o output.wav "Your text here".At runtime, the wrapper reads SHERPA_ONNX_RUNTIME_DIR and SHERPA_ONNX_MODEL_DIR to locate the engine and model, then synthesizes speech entirely locally.The result is saved as a .wav audio file at the path you specify — no network requests involved at any step.

Fully Offline — All inference runs locally; no API keys, no internet connection required after install.Cross-Platform — Pre-built runtimes available for macOS (universal), Linux x64, and Windows x64.Swappable Voice Models — Use the bundled en_US-lessac-high Piper model or replace it with any compatible sherpa-onnx TTS model for different voices or languages.Simple CLI Interface — Synthesize speech with a single command; output directly to a .wav file.Flexible Model Configuration — Override the model file, tokens file, or data directory via environment variables (SHERPA_ONNX_MODEL_FILE) or CLI flags (--model-file, --tokens-file, --data-dir).Privacy-First — Text never leaves your machine, making it suitable for sensitive or air-gapped environments.

sherpa-onnx Runtime Directory (SHERPA_ONNX_RUNTIME_DIR) — (Required) Path to the extracted sherpa-onnx binary runtime for your OS. Downloaded automatically during install.sherpa-onnx Model Directory (SHERPA_ONNX_MODEL_DIR) — (Required) Path to the extracted VITS voice model directory (e.g., vits-piper-en_US-lessac-high). Downloaded automatically during install. Additional models can be sourced from the sherpa-onnx tts-models releases.

Privacy-Sensitive Environments — Generate spoken audio from text without sending any data to external services, ideal for confidential documents or air-gapped systems.Offline Agent Workflows — Add TTS output to AI agent pipelines that must operate without a network connection.Local Voice Prototyping — Quickly test and iterate on voice-enabled applications using high-quality Piper VITS models without cloud costs.Batch Audio Generation — Automate the creation of .wav audio files from text content via simple CLI scripting.