Three kinds of free text-to-speech
Free text-to-speech comes in three kinds, and the right one depends on whether you only want to listen or need an audio file to publish:
| Kind | Examples in this list | Good for |
|---|---|---|
| Built-in read-aloud | Speak selection on the Mac, Narrator in Windows, Read aloud in Edge, Select to Speak on Android and Chromebooks | Listening to articles, emails and your own drafts with nothing to install |
| Open models and engines | Kokoro, Piper, eSpeak NG | Audio files you generate on your own machine, batch jobs, apps of your own |
| Desktop and hosted apps | Balabolka, ElevenLabs Free | Turning long documents into audio, or trying studio-quality voices |
To hear a short text right now in the browser, our text to speech tool reads it with the voices your system already has, and nothing is uploaded.
Licenses for the audio you make
If the audio goes into a video, a course or a product, the license matters more than the voice quality. What the official pages say:
- Kokoro is an open-weight model with 82 million parameters under Apache-2.0, and its model card says it can be deployed anywhere from production environments to personal projects.
- Piper moved to a new repository in 2025: the old one (MIT) was archived on October 6, 2025, and development continues in piper1-gpl under GPL-3.0. The repository links a separate Voices page; check what it says about the voice you pick before you publish with it.
- eSpeak NG is released under GPL version 3 or later.
- ElevenLabs lists a Commercial License as a feature of the Starter plan, not of the Free plan.
The read-aloud features built into systems and browsers are made for listening. Their help pages describe reading text aloud, not making audio files to publish.
Privacy: where your text goes
Open models such as Kokoro, Piper and eSpeak NG run on your own computer, so the text never leaves it. Hosted apps receive every sentence you convert, so keep contracts, health notes and anything personal out of them. Microsoft says Read aloud in Edge works online and offline, with only a few voices available offline, which tells you the extra voices come from the network.
The same rule applies the other way round: when you turn a recording into text, our speech to text tool uses your browser, and the free transcription tools list marks which services upload your audio.
Plan the length before you generate
Free plans count characters or credits, so it helps to know how long a script runs before you convert it. ElevenLabs counts its Free plan in 10,000 credits a month, roughly one credit per character for text-to-speech, and the pricing page lists about 10 minutes of audio for the Free plan. Paste your script into the script timer to see the speaking time at your chosen pace, then shorten it before you spend credits.
For podcasts and voice-overs you will probably edit the audio afterwards. The free audio editors list covers editors for cutting and leveling, and the free podcast hosting list covers where to publish.
Frequently asked questions
What is the easiest free text-to-speech on my computer?
The one already built in. On a Mac you turn on Speak selection under System Settings, Accessibility, Read & Speak and press Option-Esc. Windows 11 has Narrator built in, started with Ctrl, the Windows logo key and Enter, and Microsoft Edge has Read aloud in the More tools menu.
Which free text-to-speech can I use in a commercial video?
Kokoro is under Apache-2.0, and its model card says it can be deployed anywhere from production environments to personal projects. ElevenLabs lists its Commercial License from the Starter plan, not on the Free plan. For other tools, read the license of the exact voice or model you use.
Is there free text-to-speech that works offline?
Yes. Kokoro, Piper and eSpeak NG run on your own machine, and eSpeak NG can write its speech to a WAV file. Microsoft Edge's Read aloud also works offline, with fewer voices than online.
How much does the ElevenLabs Free plan include?
The pricing page lists 10,000 credits a month on the Free plan, with about one credit per character for text-to-speech, and about 10 minutes of audio. Unused credits do not roll over on the Free plan.
What happened to Piper?
The original Piper repository was archived on October 6, 2025 and is read-only. Development moved to the piper1-gpl repository, which is licensed under GPL-3.0 and installs with pip install piper-tts.