piper-si_LK-ashoka-medium

unicef/piper-si_LK-ashoka-medium

piper-si_LK-ashoka-medium is a text-to-speech model from unicef. Use it when you need text read aloud. It is set up for onnx. The card lists the license as mit.

A Piper text-to-speech voice for Sinhala, exported as ONNX.

Downloads · 30 days

Access

Public

Updated Mar 19, 2026

Repo size

64.4 MB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.onnx63.5 MB · 98%

At a glance

Task: Text-to-Speech
Library: onnx
License: mit
Access: Public
Created: Mar 19, 2026
Updated: Mar 19, 2026
SHA: 53a75778

Base models

rhasspy/piper-voices

Model card

From the Hugging Face model README

Piper TTS -- Sinhala (si_LK) medium

A Piper text-to-speech voice for Sinhala, exported as ONNX.

Samples

ආයුබෝවන්.

සිංහල භාෂාවෙන් කතා කිරීමට මෙම මාදිලිය භාවිතා කළ හැකිය.

මේ ලස්සන දිනයක් වගේ පෙනෙනවා.

අද කාලගුණය ඉතා සුන්දරයි.

Model Details


Language	Sinhala (si_LK)
Quality	medium
Base model	Fine-tuned from Hindi (`hi_IN`) via rhasspy/piper-checkpoints
Sample rate	22050 Hz
Format	ONNX

Usage

With Piper CLI

pip install piper-tts
echo 'ආයුබෝවන්.' | piper --model si_LK-ashoka-medium.onnx --output_file output.wav

With NVDA

Download both files and place them in NVDA's Piper voices directory:

si_LK-ashoka-medium.onnx
si_LK-ashoka-medium.onnx.json

Programmatic (Python)

import subprocess
text = "ආයුබෝවන්."
subprocess.run(
    ["piper", "--model", "si_LK-ashoka-medium.onnx", "--output_file", "output.wav"],
    input=text.encode(),
)

Files

File	Description
`si_LK-ashoka-medium.onnx`	ONNX model
`si_LK-ashoka-medium.onnx.json`	Piper config (phoneme map, sample rate, etc.)
`samples/`	Audio samples generated by this model

Training

Trained using the Piper EC2 Training Pipeline. Fine-tuned from the Hindi (hi/hi_IN/rohan/medium) checkpoint.

Task: Text-to-Speech
Library: onnx
License: mit
Languages: si
Created: Mar 19, 2026
Updated: Mar 19, 2026