Google Text to Speech: Free Voices, Limits & MP3 Options

The four places Google's free voices live

Most people say "Google Text to Speech" and mean four different things. Knowing which one you are using saves a lot of confusion.

1. The Android engine. This is the voice software behind Maps directions, Assistant replies, and accessibility readers — and it also powers Google Play Books' Read Aloud feature. On current phones the engine is listed as Speech Recognition & Synthesis (its Play Store name); older phones still call it Google Text-to-speech Engine. It is free, and once you download a voice, it works offline.
2. Select to Speak. This is the feature most people actually want. Turn it on under Settings > Accessibility > Select to Speak, tap the Accessibility button (or swipe up with two fingers), then tap any text on screen — a web article, a message, a document — and your phone reads it aloud. A small controller lets you pause, skip, or change speed.
3. Chrome and Google Docs read-aloud. In Chrome on Android, open the menu and tap Listen to this page — it reads the page while you keep browsing, though Chrome itself warns it isn't available on every website yet. On a computer, right-click a page, open Reading mode, and press Play. Google Docs is separate: Tools > Audio > Listen to this tab reads your document with a choice of voices, and eligible plans also get an audio summary option. Two honest limits: Docs listening is desktop-only and English-only, and it does not appear at all on free personal accounts.
4. Google Translate. Paste text, pick a language, tap Listen, and hear the translation spoken. In the app you can set the speed properly — profile icon > Settings > Speed gives you Normal, Slow, or Slower — which makes it genuinely useful for language learners checking pronunciation. On the web there is no speed menu; tapping Listen a second time simply plays it slower.

On iPhone: there is no Google engine to switch on — iOS uses its own voices. The Translate app still reads translations aloud, but anything system-wide runs through Apple's Spoken Content settings instead.

How to change the voice and make it sound better

The default voice sounds flat because it is the compact, low-storage version. The fix takes about thirty seconds:

1. Go to Settings > Accessibility > Text-to-speech output.
2. Confirm Google's engine is the preferred engine — look for Speech Recognition & Synthesis or Google Text-to-speech Engine, depending on your phone.
3. Tap the gear icon next to it, then Install voice data.
4. Pick your language and download a voice. Voices appear as variants — Voice I, Voice II, and so on — and the larger downloads are the higher-quality ones.
5. Drag the speech rate slider to whatever sounds natural to you — a touch faster than the default makes long articles easier to sit through — adjust pitch if the default sounds too high or low, and tap Play to hear the sample.

Once a voice is downloaded, it works offline. No signal on the train? Select to Speak still reads — just pick a downloaded voice rather than one that streams from the network. Samsung owners get a second engine, Samsung text-to-speech, and can pick either one in the same menu.

The Cloud version: generous, but it is a developer product

Google Cloud Text-to-Speech is a different product with the same name: an API developers wire into apps, offering 380+ voices across 75+ languages. According to Google Cloud's own pricing page, the free tier is genuinely generous — Standard and WaveNet voices include up to 4 million characters per month free, and Neural2, Studio, Polyglot, and Chirp 3: HD voices include up to 1 million characters per month free. Past that, Standard and WaveNet cost $4 per 1 million characters, Neural2 and Polyglot $16, Chirp 3: HD $30, and Studio $160.

The newest family, Gemini-TTS, plays by different rules: it has no free tier at all and is billed per token instead of per character (Gemini 2.5 Flash, for example, charges $0.50 per million text tokens in, plus $10 per million audio tokens out).

The catch is in Google's own words: "You must enable billing to use Text-to-Speech, and will be automatically charged if your usage exceeds the number of free characters allowed per month." So the Cloud API is free to try, not free to forget about — set up billing alerts before you build anything on it. You don't need code just to hear the voices, though: Google's product page runs a small demo where you type a sentence, pick a voice, and click "Speak It". Actual use in your own app still needs a Cloud project and code.

The one limit that matters: no MP3 download

Here is the honest summary. Google's consumer voices are excellent for listening: reading an article hands-free, checking a translation, or letting your phone read a document while you cook. But none of the built-in options — not Select to Speak, not Chrome, not Docs, not Translate — hands you an audio file you can put in a video, a podcast, or an ad.

One carve-out, because it matters: the Cloud API does return real audio files — MP3, WAV, and OGG — since that is what developers build with. So "Google can't give you an MP3" is only true if you can't (or don't want to) use an API.

That is the dividing line. If you only need to listen, Google already gave you everything, and you should just download a better voice and use Select to Speak. If you need a file without touching code — a voiceover for a YouTube video, narration for a presentation, audio for a course — you need a tool built for that job. Our free text to speech tool does exactly that: paste your script, pick a voice, and download the MP3 in your browser, no sign-up and no credit card. If voice quality is your main concern, our guide to the best AI voices compares what the current top voices actually sound like. And if your starting point is a document rather than typed text, our guide to apps that read PDFs aloud covers the PDF-to-audio route step by step. And if ElevenLabs is on your shortlist too, our ElevenLabs alternatives guide compares the free options side by side.

Google's options at a glance