Livev1.0 · 23 languages · OpenAI-compatible

Lifelike speech,
in one call.

A voice API for builders. We charge 0.22¢ per 1,000 characters(19 paise per minute) instead of premium-API rates. 23 languages. Real-time on a CPU, streaming sentence-by-sentence. Pay-as-you-go, billed by the second.

Get started free → Try the live demo
now playing · Meera · English · Cheerful
Hello, and a very warm welcome to LeanVoice.
50×cheaper per minute
5.5×real-time on a CPU
23languages, 6 voices
any regiondeploy close to users
Time to first word
<400ms

The first word leaves the server before you've finished reading this one — then speech streams on, sentence by sentence.

Streamed sentence-by-sentenceReal-time on a CPUNo GPU

Measured first-audio (first word) on a reference multi-core CPU at fast quality, streamed sentence-by-sentence. Your network round-trip is additional.

How it sounds

Text in. Waveform out.

Feed it a line in any of 23 languages. Lifelike speech streams back sentence-by-sentence, so playback starts before the full text is rendered.

→ feed
audio out →
One call

A drop-in for OpenAI TTS.

Already calling openai.audio.speech? Swap the base URL. Keep your code. A fraction of the bill.

curl
# Same shape. Same client. A fraction of the bill.
curl https://api.leanvoice.ai/v1/audio/speech \
   -H "Authorization: Bearer $KEY" \
   -d '{
     "model": "aayush-tts-1",
     "input": "Lifelike speech, 50 times cheaper.",
     "voice": "meera"
   }' --output speech.wav
Streaming · price per minute

Streams as it speaks.

Audio is streamed over HTTP, sentence by sentence, so playback starts before the full text is rendered. And every minute costs a fraction of premium APIs; lower bars are cheaper.

LeanVoice
$0.002
OpenAI tts-1
~$0.015
Amazon Polly Neural
~$0.016
Cartesia Sonic
~$0.036
ElevenLabs Multilingual
~$0.10

Approximate published list prices per minute of audio at default quality (per-character rates converted at ~750 chars/min). LeanVoice is ~8× cheaper than commodity neural TTS and ~50× cheaper than ElevenLabs.

Live demo

Type. Listen.

Every line below is synthesised live on the same engine the API runs. Glass slider for quality. Real voice cards. Pick one or describe your own.

Your words
Length—
First audio—
×realtime—
Voice—
Choose a voice
Create your own voice Clone from a 30-second sample. Yours, private, on-device. Request access.
Emotion · the part nobody talks about

Not just speech. Feeling.

Inline expression tags work natively. <laugh> · <breath> · <sigh>. And six distinct voice personalities, each with their own warmth.

Cheerful

Warm welcome

For onboarding, greetings, and the first moment a customer hears your brand.

Listen
Empathetic

Calm support

For complaints, refunds, and any moment the customer needs to feel heard.

Listen
Excited

Big announcement

Launches, festival sales, milestone notifications. Carries energy without overdoing it.

Listen
Apologetic

Sincere apology

When something has gone wrong. The right voice can rescue an interaction.

Listen
Cinematic

Storyteller

Audiobooks, podcast intros, trailers. Slow, deliberate, lush.

Listen
Calm

Meditative

Guided meditation, sleep apps, mindful onboarding. Designed to slow the listener down.

Listen
Use cases · where teams ship voice

Where lifelike speech actually pays off.

From real-time voice agents to ten-hour audiobooks. We ship the same engine, the same price, the same streaming pipeline.

Voice AI agents

Conversational copilots, sales bots, real-time agents. Streams sentence-by-sentence, so replies start playing while the rest is still being synthesised.

StreamingReal-time · on a CPU

IVR & phone menus

Replace robotic press-1-for-balance with voices that sound human. Twenty-three languages, every major accent included.

23 langsBanking · telecom · public

Audiobooks & narration

Ten-hour books for the cost of lunch. Studio quality, multi-voice character switches, inline expression tags.

10 hrsLong-form · multi-voice

E-learning & courseware

Scale lessons across every language you ship in. Refresh content monthly without re-recording a thing.

0 retakesEdTech · localisation

Podcasts & media

Intros, outros, ad-reads, fully AI-generated shows. Six distinct voice personalities, each with their own warmth.

6 voicesContent · syndication

Accessibility

Screen readers, assistive devices, civic and government services. Multi-language at the price of a legacy TTS engine.

WCAG-readyInclusive · multilingual

Notifications & alerts

OTP voice calls, payment reminders, delivery updates, fraud alerts. Fractions of a cent per notification, at any scale.

$0.001High volume · scheduled

Video & dubbing

Auto-dub a YouTube channel into ten languages, voice an ad campaign in a day, or generate the audio track for a localised explainer without a studio.

10 langsLocalisation · creators
Scale · numbers nobody else publishes

Built for insane volume.

One eight-core box serves a small contact centre. Stack them, and we'll happily serve a national bank.

StreamingAudio starts before the text endsSentence-by-sentence over HTTP, so playback begins early
23 langsOne model, every marketEnglish, Hindi, Spanish, French, German and 18 more
6 voicesDistinct personalitiesSwitch mid-call, blend across characters, no extra fee
5.5× realtimeSynthesis speedRender a one-hour audiobook in about eleven minutes
StreamingSentence by sentenceREST streams over HTTP today; a WebSocket API is on the roadmap
10k charsFree every monthGenerous free tier, no credit card, no time bomb
99.9%Uptime targetMulti-region failover and redundant edges rolling out
ComplianceEnterprise-readySOC 2 in progress; GDPR-aligned, on EU/US infrastructure
Pricing

One price. Pay only for what you use.

No plans, no slabs, no monthly minimums. Billed per second, every second of audio you generate.

0.22¢/ 1k chars
about $0.12/ hour of audio
Per minute of audio$0.0028× cheaper than Polly, 50× cheaper than ElevenLabs
Includes all 23 languages and every voice. About 50× cheaper than ElevenLabs Multilingual, 18× cheaper than Cartesia Sonic, 8× cheaper than Amazon Polly Neural. First 10,000 characters free every month; three dollars buys about 1,500 minutes of speech.
Get started free → See full pricing
Start free

Get your API key.

Create a free account and copy your key straight from the dashboard — no card, no waitlist.

Create your free account → Log in
Free tier · 10,000 characters / month · every voice & language