murmr
Powered by Qwen3-TTS

The TTS API that
speaks your language.

Describe a voice. Hear it speak. Ship it tonight.

Describe your voice
0/150
--:--
Get StartedSee Pricing

Free tier included. No credit card required.

Ship voice features in 5 minutes

If you've used OpenAI's TTS, you already know our API.

curl https://api.murmr.dev/v1/voices/design/stream \
  -H "Authorization: Bearer $MURMR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Ship voice features in five minutes.",
    "voice_description": "A calm, confident narrator",
    "language": "English"
  }'
10
languages
Infinite
voices
#1
in 9/12 multilingual benchmarks*

Qwen3-TTS leads the benchmarks

Independent evaluations show Qwen3-TTS outperforms ElevenLabs, OpenAI, and other leading models in 9 of 12 cross-lingual tasks — including naturalness, speaker similarity, and pronunciation accuracy.

GitHubPaper

Why developers choose murmr

State-of-the-Art Model

Qwen3-TTS was trained on 5 million hours of speech. Your users can tell the difference, even if they can't explain why.

Real-time Streaming

Your voice agents can finally keep up with the conversation.

10 Languages, Native Quality

Your Brazilian customers will notice. So will your German ones.

ENESPTDEFRJAKOZHRUIT

Style Instructions

Describe the voice you want in plain English. Excited narrator? Calm meditation guide? The model figures out the rest.

OpenAI-Compatible API

Point your existing OpenAI SDK at our endpoint. That's it. Your code doesn't know the difference.

Pricing That Scales With You

Start free, upgrade when you need to.

Portable Voice Embeddings

Extract voice data, store it in your own database, and use it anywhere. The recommended pattern for multi-tenant apps.

Learn more →

Stop overpaying for voice

See how murmr stacks up against the competition

murmrElevenLabsOpenAIGoogle TTS
Price$10/mo$22/moPay-as-you-goPay-as-you-go
Included chars1M100K——
Effective $/1M$10$220$15$4-16
Custom voicesAny voice from textYour voice onlyPresets onlyPresets only
OpenAI-compatible✓✗✓✗
Start free with 50K characters/month

Built for builders

See what developers are shipping with murmr

Voice Agents & Chatbots

Natural conversations in any language with style control

Game Dialogue

Dynamic NPC voices without expensive recording sessions

Content Localization

Dub videos and podcasts into 10 languages automatically

Audiobooks & Podcasts

Long-form narration that doesn't fatigue listeners

E-Learning

Accessible courses with natural pronunciation across languages

Accessibility

Screen readers and assistive tools that sound human

Pricing that makes sense

Start free. Scale when you need to.

Free

$0

50K characters/month

  • 3 saved voices
  • 10 languages
  • OpenAI-compatible API
Best Value

Starter

$10/mo

1M characters/month

  • 10 saved voices
  • Commercial license
  • Email support

Then $15/1M

Pro

$25/mo

3M characters/month

  • 25 saved voices
  • Commercial license
  • Email support

Then $12/1M

Realtime

$49/mo

3M characters/month

  • 50 saved voices
  • WebSocket access
  • For voice agents

Then $12/1M

See all plans

Install the SDK

Node.js
npm install @murmr/sdk
Python
pip install murmr
REST
curl api.murmr.dev/v1/...

Ready to ship voice?

Start free. No credit card required. Upgrade when you're ready.

Get API KeyRead the Docs

Frequently asked questions

How does murmr compare to ElevenLabs?
murmr is built on Qwen3-TTS, which independently benchmarks above ElevenLabs in 9 of 12 cross-lingual tasks. We're ~95% cheaper per million characters ($10 vs $220), built on an open-source model (no vendor lock-in), and provide an OpenAI-compatible API for easy migration.
What makes your multilingual quality better?
Most TTS APIs are English-first with other languages as afterthoughts. Qwen3-TTS was trained on 5M+ hours of multilingual data, with native speaker quality across all 10 supported languages. Your Spanish content won't sound like an English speaker reading a script.
What is VoiceDesign?
VoiceDesign lets you create any voice by describing it in plain language — 'a warm British narrator with a calm pace' or 'an energetic young woman with slight French accent.' No audio samples needed. It works in all 10 supported languages.
How fast is generation?
With streaming enabled, audio playback begins almost immediately — no waiting for the full file to generate. For voice agent applications, our WebSocket endpoint provides low-latency real-time streaming. Total generation time depends on text length.
How do I migrate from OpenAI TTS?
Swap one line of code. Our API uses the exact same endpoint structure and request format as OpenAI's TTS API. Point your existing SDK to our endpoint and you're done.
Is the free tier actually usable?
Yes. 50,000 characters/month is enough for real prototyping — roughly 30 minutes of audio. VoiceDesign, all languages, streaming, OpenAI-compatible API. No credit card required.
Can I save voices for reuse?
Yes! Design a voice once with VoiceDesign, save it, and reuse it across all your projects. Free accounts can save 3 voices, Starter gets 10, Pro gets 25, and Scale gets 100.
Can I use murmr for commercial projects?
The free tier is for personal and evaluation use. Commercial use requires a Starter plan ($10/mo) or higher. This includes commercial games, apps, videos, and client work.