orksay / README.md

Koosha H

orksay voice model

b7a7b62 3 days ago

1.56 kB

---
license: cc-by-nc-4.0
language:
  - en
pipeline_tag: text-to-speech
library_name: piper
---
# Orksay Peon voice

An English Piper text-to-speech voice for Orksay, inspired by the Warcraft III Orc Peon. The model runs locally on CPU; Orksay applies the included equalization filter after synthesis.

## Files

- `model.onnx` — Piper voice model.
- `model.onnx.json` — Piper configuration and inference settings.
- `eq.npy` — FIR equalization filter used by Orksay. Piper alone does not apply this filter.

The voice uses the `en-us` eSpeak phoneme frontend and produces 22,050 Hz audio. The ONNX file is stored with Git LFS.

## Use with Orksay

Install the speech runtime from [Orksay](https://gitlab.com/hkoosha/orksay), then copy the three voice files into its data directory:

```sh
voice_dir="${XDG_DATA_HOME:-$HOME/.local/share}/orksay/voice"
mkdir -p "$voice_dir"
cp model.onnx model.onnx.json eq.npy "$voice_dir/"
```

Run these commands from a checkout of this model repository. In the Orksay project directory, run:

```sh
bash ./install-speech.sh
./orksay say "Ready to work!" --output /tmp/orksay.wav
```

Playback requires PipeWire's `pw-play`; saving a WAV does not. Orksay loads the three files from the same directory and applies `eq.npy` after Piper synthesis.

## Peon

This voice is derived from Warcraft III recordings. The [Peon audio pack](https://github.com/PeonPing/og-packs/tree/5d1245fe0188c8da775ca8875c32ee7bf8d92c57/peon) is declared CC-BY-NC-4.0. The training script can be found in the linked orksay git repo above.

	---
	license: cc-by-nc-4.0
	language:
	- en
	pipeline_tag: text-to-speech
	library_name: piper
	---

	# Orksay Peon voice

	An English Piper text-to-speech voice for Orksay, inspired by the Warcraft III Orc Peon. The model runs locally on CPU; Orksay applies the included equalization filter after synthesis.

	## Files

	- `model.onnx` — Piper voice model.
	- `model.onnx.json` — Piper configuration and inference settings.
	- `eq.npy` — FIR equalization filter used by Orksay. Piper alone does not apply this filter.

	The voice uses the `en-us` eSpeak phoneme frontend and produces 22,050 Hz audio. The ONNX file is stored with Git LFS.

	## Use with Orksay

	Install the speech runtime from [Orksay](https://gitlab.com/hkoosha/orksay), then copy the three voice files into its data directory:

	```sh
	voice_dir="${XDG_DATA_HOME:-$HOME/.local/share}/orksay/voice"
	mkdir -p "$voice_dir"
	cp model.onnx model.onnx.json eq.npy "$voice_dir/"
	```

	Run these commands from a checkout of this model repository. In the Orksay project directory, run:

	```sh
	bash ./install-speech.sh
	./orksay say "Ready to work!" --output /tmp/orksay.wav
	```

	Playback requires PipeWire's `pw-play`; saving a WAV does not. Orksay loads the three files from the same directory and applies `eq.npy` after Piper synthesis.

	## Peon

	This voice is derived from Warcraft III recordings. The [Peon audio pack](https://github.com/PeonPing/og-packs/tree/5d1245fe0188c8da775ca8875c32ee7bf8d92c57/peon) is declared CC-BY-NC-4.0. The training script can be found in the linked orksay git repo above.