Voices and Read-Aloud

Recommendation

If you want the best Bible read-aloud experience in Logos AI, start with Kokoro.

It sounds much better than the basic system voices, works well for longer Scripture reading, and fits the product direction better than treating read-aloud as a throwaway accessibility extra.

Detection order

Logos AI checks for voices in this order:

Piper
Kokoro
macOS say
Windows built-in speech
Linux espeak-ng, espeak, or spd-say

That means the desktop app and the TUI will still read aloud even when Piper or Kokoro are missing, as long as a basic system voice is present.

Best-quality read-aloud is currently English-first because Kokoro is the main recommended path today. Other languages can still work through matching Piper voices or the operating system’s installed voices on macOS, Windows, or Linux.

Install Kokoro

Kokoro is the recommended path.

Install Python 3.11+.
Install the upstream package:

python3 -m pip install -U kokoro-onnx soundfile

Create the model directory Logos AI expects:

mkdir -p ~/.local/share/kokoro ~/.local/bin

Download these two files into ~/.local/share/kokoro/:

kokoro-v1.0.onnx
voices-v1.0.bin

Create the kokoro-speak wrapper command Logos AI looks for:

#!/usr/bin/env python3
import os
import sys
import numpy as np
from kokoro_onnx import Kokoro

home = os.path.expanduser("~")
model = home + "/.local/share/kokoro/kokoro-v1.0.onnx"
voices = home + "/.local/share/kokoro/voices-v1.0.bin"

voice = sys.argv[1] if len(sys.argv) > 1 else "am_michael"
speed = float(sys.argv[2]) if len(sys.argv) > 2 else 0.9
text = sys.stdin.read().strip()
if not text:
    sys.exit(0)

k = Kokoro(model, voices)
samples, rate = k.create(text, voice=voice, speed=speed, lang="en-us")
pcm = (np.clip(samples, -1.0, 1.0) * 32767).astype(np.int16)
sys.stdout.buffer.write(pcm.tobytes())

Save that as ~/.local/bin/kokoro-speak, then:

chmod +x ~/.local/bin/kokoro-speak

If ~/.local/bin is already on your PATH, Logos AI should detect Kokoro automatically the next time the app starts.

If you would rather not do this yourself, Kokoro or Piper setup can also be handled as paid setup help.

Install Piper

Piper is still a strong local neural option and remains a good fallback when you want a simpler setup or already have Piper models on disk.

Install the upstream package:

python3 -m pip install -U piper-tts

Put your Piper voice files in ~/.local/share/piper/.

You want the .onnx model file and usually the matching .onnx.json file beside it.

If your Python install does not expose a piper command on PATH, create this wrapper:

#!/bin/sh
exec python3 -m piper "$@"

Save it as ~/.local/bin/piper and make it executable:

chmod +x ~/.local/bin/piper

Point Logos AI at the model you want by setting PIPER_MODEL.

For the desktop app, the easiest place is ~/.config/logos/.env:

PIPER_MODEL=/Users/yourname/.local/share/piper/en_US-lessac-medium.onnx

The Wails desktop app also reads ~/.logos.env if that fits your setup better.

Setup help and future automation

Voice setup is also a good fit for a paid convenience layer today and a future premium automation script later.

The practical idea is:

self-serve docs stay available
paid setup help can handle Kokoro or Piper for people who want it done for them
a future premium add-on can automate more of the local install on supported systems

Basic system fallback

If Kokoro and Piper are missing, Logos AI now falls back cleanly instead of disabling read-aloud.

macOS

Uses the built-in say command automatically.

Windows

Uses Windows built-in speech through PowerShell and the installed system voices.

Linux

Uses espeak-ng, espeak, or spd-say when one of them is installed.

For Ubuntu or Debian systems, this is the simplest fallback install:

sudo apt install espeak-ng speech-dispatcher

Official upstream references

Keep install commands handy

Once voices are configured, the app install itself is still just:

brew tap jd4rider/logos-ai https://github.com/jd4rider/homebrew-logos-ai
brew install --cask logos-ai

scoop bucket add logos-ai https://github.com/jd4rider/scoop-logos-ai
scoop install logos-ai

curl -fsSL https://raw.githubusercontent.com/jd4rider/logos-releases/main/install.sh | bash

Back to install guide