How to Get Better Text-to-Speech Voices for Your Target Language

When you tap the speaker icon on a phrase in Polyglotton, your browser uses your device’s built-in text-to-speech engine to read it aloud. On most devices this works well for your native language — but sounds robotic or fails entirely for the language you are trying to learn.

This guide explains why that happens and shows you exactly how to fix it on each platform.

Why Your Device Doesn’t Have the Voice You Need

Your operating system was configured at the factory for the region it was sold in. Installing every language voice up front would add 200 MB to 2 GB per language — and with roughly 100 supported languages across Windows, macOS, and Android, that quickly becomes 20–200 GB of audio data most users will never use.

So manufacturers ship only the voices for your locale and let you download the rest on demand.

The good news: this is quietly changing. The new generation of AI-synthesised voices — Apple’s Neural TTS in macOS Sonoma and later, Microsoft’s neural voices in Windows 11, and Google’s WaveNet voices on Android — are dramatically smaller and better-sounding than the older concatenative models. As on-device AI becomes standard (Apple Intelligence in iOS/macOS 26, Google Gemini Nano on Pixel, Samsung Galaxy AI), your OS will increasingly be able to generate any voice on the fly rather than pre-downloading recordings. We are probably only a year or two away from every language sounding natural out of the box. For now, a quick manual install gets you there.

macOS (Ventura, Sonoma, Sequoia)

macOS bundles Apple’s highest-quality “Enhanced” and “Premium” voices, but only your system language is downloaded by default.

Open Apple menu → System Settings
Click Accessibility in the sidebar
Click Spoken Content
Next to System voice, click the dropdown and choose Manage Voices… (Sonoma/Sequoia) — or click the ⓘ info icon next to the selected voice (Ventura)
In the voice list, find your target language (e.g. Hungarian, Dutch, French)
Expand the language to see available voices — look for Enhanced or Premium quality options, they sound noticeably more natural
Click the ⬇ download button next to the voice you want
Wait for the download to complete, then restart your browser

After restarting, the new voice will appear in Polyglotton’s audio settings panel under the voice picker for that language.

Tip for macOS Sequoia: You can also speak-type “Add a voice for [language]” to Siri and it will open the right settings panel directly.

Windows 11

Windows 11 uses Microsoft’s neural voices (called Natural voices) which sound very lifelike — but again, only your display language is installed by default.

Method 1 — through Language Settings (recommended)

Open Settings → Time & language → Language & region
Click Add a language and search for your target language (e.g. French, German, Hungarian)
During installation, make sure Text-to-speech is checked in the optional features list
Click Install and wait for the download to finish
Restart your browser to make the new voice available

Method 2 — through Speech Settings

Open Settings → Time & language → Speech
Under Manage voices, click Add voices
Select your target language and click Add
Restart your browser after installation

Note: Windows may install a basic voice first. For the Natural (neural) version, go back to Settings → Accessibility → Narrator → Add natural voices after the language pack is installed.

Android

Android supports two main TTS engines. Most stock Android phones use Google Text-to-Speech; Samsung Galaxy devices also bundle the Samsung TTS engine. Either way, the install steps are similar.

Google Pixel and stock Android

Open Settings → Accessibility → Text-to-speech output
Tap the ⚙ gear icon next to Google Text-to-speech
Tap Install voice data
Find your target language in the list and tap the ⬇ download icon next to it
Once downloaded, go back and set that language as your preferred language if needed

Samsung Galaxy

Samsung Galaxy phones have their own settings path:

Open Settings → General Management → Text-to-speech
(On some models: Settings → Accessibility → Text-to-speech)
Under Preferred engine, choose Google Text-to-speech for the widest language coverage, or keep the Samsung engine if your language is listed
Tap the ⚙ gear icon next to the engine
Tap Install voice data
Select your target language and tap ⬇ download
Close and reopen your browser (Chrome or Samsung Internet) to detect the new voice

Samsung Internet vs Chrome: Both use the system TTS engine, but Chrome on Android tends to have better Web Speech API compatibility. If voices are still missing, try switching browsers.

iOS and iPadOS

iOS uses Apple’s system voices, which are managed the same way as macOS.

Open Settings → Accessibility → Spoken Content → Voices
Tap your target language in the list
Tap the ⬇ download icon next to the voice you want (choose Enhanced for the best quality)
Reload Safari after installation

Voice playback in Polyglotton uses the Web Speech API, which is fully supported in Safari on iOS 16+ and Chrome on Android.

After Installing: Select the Voice in Polyglotton

Once you have downloaded a voice, open the audio settings panel in Polyglotton (the speaker button at the bottom right of a lesson page) and select the new voice from the Voice dropdown — it will appear automatically once your browser detects it.

You can also adjust Speed and Pitch to match the pace that feels natural to you. Slower speeds are great for catching pronunciation details; higher speeds help build fluency once you are comfortable with the phrases.

What’s Coming

The next wave of on-device AI is making this whole manual process obsolete. Apple Intelligence (rolling out across iOS 26 and macOS Tahoe in 2025–2026) includes a neural synthesis layer that can generate natural-sounding speech in any supported language directly on your device — no pre-downloaded voice pack required. Microsoft is taking a similar path with Copilot-powered neural voices in Windows, and Google Gemini Nano on Android already handles multilingual synthesis offline.

Within the next year or two, your device should be able to speak any language as naturally as your native one — automatically, without any setup. Until then, the steps above get you most of the way there today.