Voice library

106 AI voices. Listen to every one.

Every sample below is real output from the generator - native speakers of fourteen languages, six universal voices and thirteen OpenAI voices. Press play, find yours, then paste your text - Speecho detects the language and suggests matching voices.

English voices

10 native speakers

Three American and three British voices - pick your English accent →

Spanish voices

8 native speakers

More about Spanish text to speech →

German voices

8 native speakers

More about German text to speech →

French voices

7 native speakers

More about French text to speech →

Portuguese voices

7 native speakers

More about Portuguese text to speech →

Italian voices

7 native speakers

More about Italian text to speech →

Japanese voices

8 native speakers

More about Japanese text to speech →

Arabic voices

7 native speakers

More about Arabic text to speech →

Korean voices

4 native speakers

More about Korean text to speech →

Turkish voices

4 native speakers

More about Turkish text to speech →

Polish voices

4 native speakers

More about Polish text to speech →

Dutch voices

5 native speakers

More about Dutch text to speech →

Ukrainian voices

4 native speakers

More about Ukrainian text to speech →

Vietnamese voices

4 native speakers

More about Vietnamese text to speech →

Universal voices

6 voices · ElevenLabs

These read any of 29+ languages - with an English accent. Great for English content and for languages without a native voice yet; for the fourteen languages above, the natives will sound better.

Aria Female
Sarah Female
Laura Female
George Male
Charlie Male
Callum Male

OpenAI voices

13 voices · Standard tier

The budget tier: 50 credits per 1,000 characters, 50+ languages, and the only tier with delivery styles. Marin and Cedar are OpenAI's newest and most natural.

Marin Female
Cedar Male
Alloy Neutral
Ash Male
Ballad Male
Coral Female
Echo Male
Fable Neutral
Nova Female
Onyx Male
Sage Neutral
Shimmer Female
Verse Male

Delivery styles

OpenAI voices

The same voice can read the same text six different ways - pick a style and the tone, pace and emphasis follow. ElevenLabs voices read in their natural delivery.

DefaultThe voice as it naturally reads.
NarrationUnhurried and even, like an audiobook narrator.
UpbeatBright and energetic, promo-style delivery.
CalmSlow and soothing, generous pauses.
NewsMeasured, crisp, broadcast-neutral.
ConversationalRelaxed, like talking to a friend.

Heard one you like?

Paste your text, preview free, pay per character - from a one-time credit pack that never expires. No subscription.

Try text to speech →

150 free credits on email confirmation · how pricing works

Frequently asked questions

What is the difference between native and universal voices?

A universal voice can read any language, but it keeps its own English accent - German read by Aria sounds like an English speaker reading German. A native voice is a speaker of that language: Susi reading German sounds German. Speecho ships native speakers for fourteen languages: English (US and UK), Spanish (Latin American and Castilian), German, French, Portuguese, Italian, Japanese, Arabic, Korean, Dutch, Polish, Turkish, Ukrainian and Vietnamese.

Which quality tier does each voice belong to?

The 13 OpenAI voices run on the Standard tier (50 credits per 1,000 characters). The 93 ElevenLabs voices - native and universal - are shared by the Enhanced tier (90 credits, Flash model) and the Premium tier (180 credits, Multilingual v2, the most natural). Same voices, better model as you go up. The one exception is Vietnamese, which Multilingual v2 cannot read: for Vietnamese, Enhanced is the top tier.

Are these samples real?

Yes - every sample on this page was generated by the same models and voices the app uses, with no post-processing. What you hear is what your text will sound like.

Can I download the audio as MP3?

Yes. Generated clips play in the browser and download as MP3. They stay in your history for 90 days, and you can delete them at any time.

Can I use the generated audio commercially?

Yes. Generation runs on paid API tiers of OpenAI and ElevenLabs, whose licenses permit commercial use of the generated audio. All voices are licensed stock voices - Speecho does not clone voices.

How much does it cost?

You pay per 1,000 characters of text from a prepaid credit balance - no subscription, and credits never expire. Confirming your email gives you 150 free credits, enough to voice about 3,000 characters on the Standard tier.