🔥 Black Friday AI API deals are live· New drops every MondayEnds in--d --:--:--See deals →
Models/Audio/Text to Speech
Live11 models

Text to Speech API — Natural AI Voice Generation

Muapi's text-to-speech API converts written text into natural, human-like audio through one unified REST integration. Choose from MiniMax, Google Gemini, ElevenLabs (including the Eleven v4 API), and Fish Audio (the Fish Audio S2.1 Pro API) for single-voice narration or multi-speaker dialogue.Use studio-grade or turbo-fast generation with controls for voice, emotion, pace, stability, and similarity. The same asynchronous submit-and-poll pattern works across every model, with pricing from $0.015 per 1,000 characters.

i

Text to speech is live on Muapi

Choose a single-voice or multi-speaker model, submit your script, and poll the returned request ID for a downloadable audio file.

MiniMax Speech

Single-voice narration with studio and turbo tiers2 models
TTSLive
MiniMax

MiniMax Speech 2.6 HD

Studio-quality speech with clear pronunciation, smooth pacing, and realistic emotion for audiobooks, podcasts, and narration.

Text → audio
$0.13/1,000 chars
Try Model
TTSLive
MiniMax

MiniMax Speech 2.6 Turbo

Fast, lightweight speech with natural quality and adjustable speed, volume, pitch, and emotion for quick-turnaround audio.

Text → audioFast
$0.07/1,000 chars
Try Model

Gemini TTS

Expressive multi-speaker dialogue with voice and tone controls4 models
TTSLive
Google Gemini

Gemini 3.1 Flash TTS

Cost-efficient multi-speaker speech with distinct voices, accents, emotional styles, and inline tone tags such as [whispers].

Dialogue → audioFast
$0.035/1,000 chars
Try Model
TTSLive
Google Gemini

Gemini 2.5 Pro TTS

Premium multi-speaker speech for audiobooks, cinematic dialogue, and long-form narration with high-fidelity delivery.

Dialogue → audio
$0.035/1,000 chars
Try Model
TTSnew
Google Gemini

Dialogue → audioFast
$0.015/1,000 chars
Try Model
TTSnew
Google Gemini

Dialogue → audioFast
$0.01/1,000 chars
Try Model

ElevenLabs

Eleven v4 expressive speech, fast single-voice speech, and multilingual dialogue4 models
TTSLive
ElevenLabs

ElevenLabs TTS Turbo 2.5

Fast single-voice speech with tunable stability, similarity, and speaking rate for voiceovers, IVR, and high-throughput narration.

Text → audioFast
$0.05/1,000 chars
Try Model
TTSLive
ElevenLabs

ElevenLabs Text to Dialogue V3

Multilingual multi-speaker dialogue from a structured script, with per-speaker voice IDs and lifelike turn-taking.

Dialogue → audio
$0.10/1,000 chars
Try Model
TTSnew
ElevenLabs

Eleven v4

ElevenLabs' Eleven v4 text-to-speech for expressive single-voice narration: up to 2,000 characters per request at $0.08/1,000 characters, returned as an MP3.

Text → audio
$0.08/1,000 chars
Try Model
TTSnew
ElevenLabs

Eleven v4 Turbo

ElevenLabs' Eleven v4 Turbo text-to-speech for fast, low-cost single-voice narration: up to 5,000 characters per request at $0.04/1,000 characters, returned as an MP3.

Text → audioFast
$0.04/1,000 chars
Try Model

Fish Audio S2.1 Pro

Long-form narration up to 10,000 characters per request1 models
TTSnew
Fish Audio

Fish Audio S2.1 Pro

Long-form text-to-speech: up to 10,000 characters per request, billed at $28 per 1M UTF-8 bytes (about $0.028/1,000 English characters), returned as an MP3.

Text → audio
$28/1M UTF-8 bytes
Try Model

What is a Text to Speech API?

A text-to-speech (TTS) API converts written text into spoken audio through a single HTTP request. Submit a script, choose a voice and delivery style, and receive a downloadable audio file without building an audio-engineering pipeline.

Muapi exposes the live TTS endpoints through the same unified REST pattern: one API key, one request and poll flow, and pay-as-you-go billing. Choose single-voice narration for voiceovers and IVR (including the Eleven v4 API and the Fish Audio S2.1 Pro API), or multi-speaker dialogue for characters, podcasts, and audiobooks.

Text to speech capabilities

Multi-speaker dialogue

Gemini TTS and ElevenLabs Text to Dialogue V3 assign distinct voices, accents, and delivery styles to scripted conversation turns.

Inline emotion and tone

Gemini models support inline tags such as [whispers], [shouting], and [determination] for expressive control within a line.

Studio and turbo tiers

Use MiniMax Speech 2.6 HD or Gemini 2.5 Pro TTS for fidelity, or their faster counterparts when latency and cost matter more.

Tunable voice parameters

Adjust speed, volume, pitch, emotion, stability, similarity, and other model-specific controls.

Custom and cloned voices

ElevenLabs endpoints accept a custom voice ID. Create a reusable voice first with a compatible voice-cloning workflow.

Four providers, one API

MiniMax, Google Gemini, ElevenLabs, and Fish Audio models share Muapi authentication, asynchronous tasks, result polling, and billing.

Text to speech use cases

Narration and voiceovers

Produce audiobook chapters, video narration, ads, explainers, and social clips from scripts.

Characters and podcasts

Render multi-host conversations and character dialogue with distinct voices and emotional direction.

IVR and app audio

Generate phone prompts, notifications, accessibility audio, and other dynamic speech at scale.

Localization

Use multilingual dialogue and language controls for global content and learning materials.

Text to speech model comparison, including the Eleven v4 API and Fish Audio S2.1 Pro API

ModelProviderPriceBest For
MiniMax Speech 2.6 HDMiniMax$0.13/1,000 charactersStudio-quality single-voice narration
MiniMax Speech 2.6 TurboMiniMax$0.07/1,000 charactersFast single-voice generation
ElevenLabs TTS Turbo 2.5ElevenLabs$0.05/1,000 charactersHigh-throughput narration and IVR
Gemini 3.1 Flash TTSGoogle$0.035/1,000 charactersFast multi-speaker dialogue
Gemini 2.5 Pro TTSGoogle$0.035/1,000 charactersStudio-quality multi-speaker dialogue
ElevenLabs Text to Dialogue V3ElevenLabs$0.10/1,000 charactersMultilingual multi-speaker conversations
Eleven v4ElevenLabs$0.08/1,000 charactersExpressive single-voice narration, up to 2,000 characters per request
Fish Audio S2.1 ProFish Audio$28/1M UTF-8 bytes (~$0.028/1,000 English characters)Long-form narration, up to 10,000 characters per request

Fish Audio S2.1 Pro is billed per UTF-8 byte: English text is about one byte per character, while non-Latin scripts use three to four bytes per character.

Text to speech API examples

Submit plain text or structured dialogue with your Muapi API key, save the request ID, and poll the standard prediction result endpoint for the audio URL.

Generate single-voice speech

curl -X POST https://api.muapi.ai/api/v1/minimax-speech-2.6-hd \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"Welcome to Muapi. Your workspace is ready.","voice_id":"Friendly_Person","speed":1,"emotion":"happy","format":"mp3"}'

# Response: {"request_id":"REQUEST_ID"}

Use a MiniMax or ElevenLabs single-voice endpoint with prompt, voice, and optional delivery parameters.

Generate Gemini dialogue

curl -X POST https://api.muapi.ai/api/v1/gemini-3-1-flash-tts \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"speakers":[{"speaker_id":"Speaker 1","voice_name":"Kore","accent":"Neutral","style":"Empathetic","pace":"Natural"},{"speaker_id":"Speaker 2","voice_name":"Puck","accent":"British (RP)","style":"Deadpan","pace":"Natural"}],"dialogue_turns":[{"speaker_id":"Speaker 1","text":"[warmly] Welcome to the show."},{"speaker_id":"Speaker 2","text":"[whispers] Let us begin."}]}'

Define speakers and ordered dialogue turns; each turn references the voice configuration it should use.

Generate speech with the Eleven v4 API

curl -X POST https://api.muapi.ai/api/v1/elevenlabs-v4 \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"Welcome to the future of voice. This is Eleven v4, speaking with natural rhythm and emotion.","voice_id":"voice_daniel"}'

# Response: {"request_id":"REQUEST_ID"}

Send up to 2,000 characters in prompt and choose voice_daniel or voice_liam; Eleven v4 is $0.08 per 1,000 characters.

Generate long-form speech with the Fish Audio S2.1 Pro API

curl -X POST https://api.muapi.ai/api/v1/fish-s2-1-pro \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"Welcome to the show. Today we look at how voice models have changed the way we tell stories.","voice_id":"voice_fish_narrator"}'

# Response: {"request_id":"REQUEST_ID"}

Send up to 10,000 characters in prompt with voice_fish_narrator; billed at $28 per 1M UTF-8 bytes.

Generate ElevenLabs dialogue

curl -X POST https://api.muapi.ai/api/v1/elevenlabs-text-to-dialogue-v3 \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"dialogue":[{"text":"Welcome to the show.","voice_id":"ZQe5CZNOzWyzPSCn5a3c"},{"text":"Let us begin.","voice_id":"EXAVITQu4vr4xnSDxMaL"}],"language_code":"en","stability":0.5}'

Pass a structured dialogue array with a voice ID per turn for multilingual multi-speaker output.

Poll the audio result

curl https://api.muapi.ai/api/v1/predictions/REQUEST_ID/result \
  -H "x-api-key: YOUR_API_KEY"

# Read the generated audio URL after status becomes completed.

Poll until status is completed, then read the downloadable audio URL from the response.

Text to speech API FAQ

What is the Text to Speech API?

Muapi's TTS API converts written text into natural audio through MiniMax, Google Gemini, ElevenLabs, and Fish Audio endpoints for single-voice narration and multi-speaker dialogue.

Which model should I use for multi-speaker dialogue?

Use Gemini 3.1 Flash TTS or Gemini 2.5 Pro TTS for speaker profiles and inline tone tags, or ElevenLabs Text to Dialogue V3 for multilingual structured conversations with per-speaker voice IDs.

How much does text-to-speech cost?

Pricing starts at $0.028 per 1,000 English characters for Fish Audio S2.1 Pro (billed at $28 per 1M UTF-8 bytes), $0.035 per 1,000 characters for Gemini 3.1 Flash and 2.5 Pro TTS, $0.05 for ElevenLabs TTS Turbo 2.5, $0.08 for Eleven v4, $0.10 for ElevenLabs Dialogue V3, $0.07 for MiniMax Turbo, and $0.13 for MiniMax HD. Gemini 3.8 Flash and Flash Lite are $0.015 and $0.01 per 1,000 characters.

How do I use the Eleven v4 API and what is its length limit?

Call the elevenlabs-v4 endpoint with prompt (up to 2,000 characters) and a voice_id of voice_daniel or voice_liam, then poll the request ID for an MP3. Eleven v4 costs $0.08 per 1,000 characters.

What is the Fish Audio S2.1 Pro API?

The Fish Audio S2.1 Pro API (fish-s2-1-pro) is a long-form text-to-speech endpoint that accepts up to 10,000 characters per request with voice_fish_narrator and returns an MP3. It is billed at $28 per 1M UTF-8 bytes, which is about $0.028 per 1,000 English characters.

Can I use a custom or cloned voice?

Yes. ElevenLabs endpoints accept a custom ElevenLabs voice ID, and MiniMax supports trained voice IDs where the selected model schema allows them.

What audio format is returned?

Each model returns a downloadable audio file URL from the standard prediction result endpoint, ready to embed, download, or pass into a video or lipsync pipeline.

How do I access Text to Speech API models?

Create a Muapi API key, call the endpoint that matches your script, and poll the returned request ID. No waitlist is required for the live models.

Ready to generate natural speech?

Use one Muapi API key for narration, character dialogue, IVR, podcasts, and multilingual audio.