MotionShineShort answer:
- A voice-over file you can use in a video, free, with nothing uploaded: the free Text to speech tool on this site. It saves MP3 or WAV with matching subtitles, and the voices run in your browser tab.
- Hear any text read aloud on a Mac, nothing to install: the Mac's own Read & Speak (called Spoken Content before macOS Tahoe 26).
- Hear web pages and PDFs read aloud in a browser: Microsoft Edge's Read aloud.
- Saving audio files on Windows with the voices you already have: Balabolka.
- The most lifelike AI voices, for personal use only: ElevenLabs' free plan, about 10 minutes a month, with credit to ElevenLabs.
- Listening to documents on your phone: NaturalReader or ElevenReader, but neither free plan saves a file.
Not for you if you need voice cloning or a voice in a language outside the nine our tool speaks: then a paid AI voice service is the honest answer.
How we checked these tools
We make MotionShine and the Text to speech tool on this site, so weigh what we say about it the way you weigh any vendor's own page. Every fact about the other tools comes from the maker's own website, pricing page or help center, checked on 7 November 2026. When a maker's pages do not say something, the table says "not stated". We did not rate voice quality, because that is a matter of taste and we did not run a listening test.
"Free" here means you can make or hear speech without paying. Where the free plan only lets you listen and not download, we say so, because that is the limit most lists leave out.
Which job do you have?
| You want to | What matters | Look at |
|---|---|---|
| Make a voice-over for a video or a course | Saving a file, commercial use, pauses | Text to speech tool, ElevenLabs (paid for commercial) |
| Listen to articles, PDFs or notes | Long text, phone apps, speed | Read & Speak on Mac, Edge Read aloud, NaturalReader, ElevenReader |
| Turn a long document into MP3 files | Saving files, chapters | Balabolka, the audiobook maker |
| Keep private text private | Where the voice runs | Text to speech tool, Mac voices, Balabolka, Piper |
| Build TTS into your own project | Open source, command line | Piper, the Mac's say command |
All 10 at a glance
| Tool | Free limit | Saves a file on the free plan | Commercial use when free | Runs on | Best for |
|---|---|---|---|---|---|
| MotionShine Text to speech | Free, 20,000 characters a run, no monthly cap | Yes: MP3, WAV, SRT, VTT | Yes, by each voice's licence; some ask for credit | Any modern browser | Voice-overs, with nothing uploaded |
Mac Read & Speak and say | Free, built in | Only with the say command (AIFF) | Not stated | Every Mac | Hearing text read aloud |
| Microsoft Edge Read aloud | Free | Not stated, listening only | Not stated | The Edge browser | Web pages and PDFs |
| NaturalReader | Unlimited Free Voices; 20,000 characters a day with Lite Voices | No: "Cannot convert to mp3" | Downloads from paid personal plans are personal use only | Web, iOS, Android, Chrome | Listening to documents |
| Balabolka | Freeware | Yes, an audio file | Depends on the voices you have | Windows | Saving files with Windows voices |
| ElevenLabs | 10,000 credits a month, about 10 minutes | Not stated for the free plan | No, personal use only, credit required | Web | Trying the most lifelike voices |
| ElevenReader | 10 hours of listening a month | No: "audio files can't be exported" | No file to use | Web, iOS, Android, Chrome | Listening on a phone |
| Speechify | 10 voices, up to 1.5x speed | Not stated | Not stated | Phone, browser and desktop apps | Reading along at speed |
| Murf | 10 minutes of voice generation | No downloads | No file to use | Web | Trying studio voices |
| Piper | Free, open source (GPL-3.0) | Yes | Check each voice's licence | Your own computer, command line | Developers and offline use |
Facts from each maker's own page, checked on 7 November 2026. Murf's free plan facts come from Murf's own article of 26 October 2026, because its pricing page did not load.
The best free text to speech software, one by one
1. MotionShine Text to speech (our tool)
Price: free, no sign-up, no watermark, no monthly character allowance. Each run takes up to 20,000 characters, which the page puts at about 20 minutes of speech.
What it does: you paste or open a script (.txt, .md, .docx or a subtitle file), pick a voice, set the speed from 0.5x to 2x and the pauses after sentences and between paragraphs, and it reads the text sentence by sentence. You can play what is ready while the rest is made, type [pause 2s] for a longer break, and add pronunciation fixes for names it says wrong. It saves MP3 at 128 or 64 kbps or WAV, plus SRT and VTT subtitles timed to the audio. There are 20 voices in nine languages: eight American and British English voices, and Chinese, Spanish, French, German, Russian, Arabic, Hindi and Bangla. The voice models (Kokoro for English and Chinese, VITS voices for the others) download to your browser the first time, about 92 MB for English, and run there, so your text is never uploaded.
Good for: voice-overs for screen recordings, tutorials and social videos, where you need a file, subtitles and the right to use the audio. Each voice's licence allows commercial use; the Hindi voices and the Bangla woman ask for credit to SYSPIN, and the Arabic voice to Open Bible.
Weak point: it is slower than a cloud service, because your computer does the work. The tool's page says it made about one second of speech every two seconds on a Mac. It has no voice cloning and no pitch control, and the Chinese, Hindi, Bangla, Russian and Arabic voices read only their own script.
For a script with two speakers, there is also multi-voice text to speech.
2. The Mac's own voices: Read & Speak and say
Price: free on every Mac.
What it does: turn on Speak selection in System Settings, Accessibility, Read & Speak (Spoken Content on macOS Sequoia 15), select text in any app and press Option-Esc. Apple's guide also lists Edit > Speech > Start Speaking in many apps. You can pick the system voice, download more voices and change the speaking rate. To save a file, the Terminal command say -o speech.aiff "Your text" writes an AIFF file, per the command's own manual page. Full steps: how to use text to speech on Mac.
Good for: proofreading by ear, listening to long articles, accessibility.
Weak point: the Accessibility setting is for listening, not for making files. Saving needs Terminal, and the result is AIFF, not MP3.
3. Microsoft Edge Read aloud
Price: free in the Edge browser.
What it does: Microsoft's support page says Read aloud reads web pages and PDFs out loud, with voice options and reading pace on its toolbar, and that you can download more voices and languages.
Good for: listening to a long article or a PDF without installing anything else.
Weak point: Microsoft's page describes listening only and does not mention saving an audio file.
4. NaturalReader
Price: a free plan; paid personal and commercial plans. We could not read the price page on the day, so the price is not stated here.
What it does: NaturalReader's help center (updated 13 July 2026) lists the free plan as unlimited use of Free Voices, 20,000 characters a day with Lite Voices, and 4,000 shared characters a day with Plus and Cloned Voices. It runs on the web, iOS and Android, and as a Chrome extension.
Good for: listening to documents and web pages on a computer and a phone.
Weak point: the free plan "Cannot convert to mp3" and has no OCR, and Free Voices cannot be turned into a download on any plan. MP3s made with the paid personal voices are for personal use only; public or business use needs the separate Commercial product.
5. Balabolka
Price: freeware. The maker says it stays free as long as people buy his other program.
What it does: Balabolka reads text with any computer voice installed on your system, through Microsoft's Speech API, and saves on-screen text as an audio file. It opens many file types, including DOCX, EPUB, PDF and HTML, and it can store the text inside MP3 tags so players show it like lyrics. A substitution list fixes how words are said. There is a portable version and a command line version.
Good for: Windows users who want audio files of long text with no account and no limit.
Weak point: it is Windows software, and it sounds only as good as the voices on your computer.
6. ElevenLabs
Price: Free, then Starter at $6 a month ($1 for the first month), which adds a commercial licence.
What it does: the free plan has 10,000 credits a month, which ElevenLabs puts at about 10 minutes of audio. Its FAQ says Eleven v4 supports 90+ languages.
Good for: hearing what current AI voices can do, for personal projects.
Weak point: ElevenLabs says the free plan "covers personal, non-commercial use" and "requires attribution to ElevenLabs", and your text goes to its servers. For a YouTube voice-over you would need a paid plan.
7. ElevenReader
Price: free plan, then Ultra at $8.25 a month billed yearly.
What it does: a reading app from ElevenLabs. The free plan gives 10 hours of text to audio every month, on the web, iOS, Android and Chrome, and syncs between them.
Good for: listening to articles and books on your phone.
Weak point: its pricing page says "audio files can't be exported from the app". Offline listening stays inside the app.
8. Speechify
Price: free plan; Premium at $29 a month or $139 billed yearly.
What it does: the free plan has 10 voices, which Speechify's own page calls "robotic sounding", and listening at up to 1.5x speed. Its site lists apps for iPhone, iPad, Android, Chrome, Edge, the web, Mac and Windows.
Good for: reading along with text at a faster pace.
Weak point: the better voices and faster speeds are on Premium, and the pricing page does not say whether the free plan saves MP3 files.
9. Murf
Price: a free plan; paid plans not read on the day (the pricing page did not load).
What it does: Murf's own article of 26 October 2026 lists its free plan as all features for 10 minutes of voice generation, for one user, with no downloads.
Good for: trying Murf's studio voices before you pay.
Weak point: with no downloads, the free plan cannot give you a voice-over file.
10. Piper
Price: free and open source, GPL-3.0.
What it does: its GitHub page calls Piper "a fast and local neural text-to-speech engine". You install it with pip and run it from the command line or from your own code, and it speaks on your own computer.
Good for: developers, home automation and anyone who wants offline speech with no account.
Weak point: there is no app to click through, and each voice model has its own licence to check before you use the audio.
Free, free plan or trial: what the word means
- Free with no allowance: the Text to speech tool, the Mac's voices, Edge Read aloud, Balabolka and Piper. You can use them as much as you like.
- Free plan with a monthly allowance: ElevenLabs (about 10 minutes), ElevenReader (10 hours of listening), NaturalReader (daily character limits per voice type), Murf (10 minutes).
- Free to listen, pay to download: NaturalReader, ElevenReader and Murf's free plans do not give you a file. That matters if you came for a voice-over.
Can you use free text to speech in a YouTube video?
It depends on the tool's licence, not on whether it is free. ElevenLabs' free plan is personal use only and asks for credit. NaturalReader's personal MP3s are for personal use only. The Text to speech tool's voices are under licences that allow commercial use, and the Hindi, Bangla woman and Arabic voices ask for credit. Whatever tool you use, also check the rules of the platform you post on about AI voices.
Where does your text go?
Cloud services (ElevenLabs, ElevenReader, NaturalReader, Speechify, Murf) send your text to their servers to make the speech. That is fine for a blog post and less fine for a contract, a patient note or an unreleased script. The Text to speech tool, the Mac's own voices, Balabolka and Piper make the speech on your own computer.
Keep a paid AI voice service if
- You need a voice clone of yourself, or many voices with emotion controls.
- You work in a language the free options here do not speak well.
- You make hours of narration a week and want it fast, on someone else's computers.
Which one should you pick?
- A voice-over for a screen recording or tutorial: the Text to speech tool, then put it under your video. MotionShine has voice over and music lanes in its editor.
- Listening on a Mac: Read & Speak.
- Listening in the browser: Edge Read aloud.
- Audio files on Windows: Balabolka.
- A book or a long PDF as MP3s: the audiobook maker or PDF to audio. See how to turn a PDF into an audiobook.
- The most lifelike voice for a personal project: ElevenLabs' free plan.
If you want your own voice on the video instead, how to record your voice on a Mac covers the free ways, and the tutorial video script template helps either way.
FAQ
Is there a completely free text to speech?
Yes. The Mac's own voices, Edge Read aloud, Balabolka, Piper and the Text to speech tool on this site cost nothing and have no monthly allowance. Of these, Balabolka, Piper and the Text to speech tool save audio files without a Terminal command. Many AI voice services also have free plans, but those come with a monthly limit, and several let you listen without downloading.
What is the most realistic free text to speech?
That is a matter of taste, and we did not run a listening test. AI voice models such as the ones behind ElevenLabs, NaturalReader's Plus voices and the Kokoro voices in our tool sound much closer to a person than older system voices. The fairest test is your own script: paste the same paragraph into two or three tools and listen.
Can I download text to speech as an MP3 for free?
Yes, with the Text to speech tool (MP3 or WAV), Balabolka and Piper. On a Mac, the say command saves AIFF, which most editors open. NaturalReader's free plan cannot convert to MP3, ElevenReader cannot export audio, and Murf's free plan has no downloads, per their own pages on 7 November 2026.
Does Windows or Mac have built-in text to speech?
Yes. On a Mac it is Read & Speak in Accessibility settings (Spoken Content on Sequoia 15), with Option-Esc to speak selected text. On Windows, Edge's Read aloud reads web pages and PDFs, and Balabolka uses the voices Windows already has.
Can I use free text to speech for commercial use?
Only if the tool's licence says so. ElevenLabs' free plan is non-commercial and needs credit. NaturalReader's personal downloads are personal use only. The voices in our Text to speech tool are under licences that allow commercial use, with credit asked for some.
Does text to speech work offline?
Some does. The Mac's voices, Balabolka and Piper run on your computer. The Text to speech tool needs the internet once to download the voice, about 92 MB for English, and then makes the speech in your browser. The cloud services need a connection every time.
