Learn more">
Skip to navigation
/n Step-by-step guide for using text to speech in ElevenLabs.\n </Card>\n\n <Card title=\"Developers\" icon=\"duotone code\" href=https://elevenlabs.io/"/docs/eleven-api/quickstart/">/n Learn how to integrate text to speech into your application.\n </Card>\n\n <Card title=\"API reference\" icon=\"duotone brackets-curly\" href=https://elevenlabs.io/"/docs/api-reference/text-to-speech/convert/">/n Full API reference for the Text to Speech endpoint.\n </Card>\n\n <Card title=\"MCP\" icon=\"duotone plug\" href=https://elevenlabs.io/"/docs/eleven-agents/operate/hosted-mcp/">/n Generate speech from Claude or any MCP client with the hosted MCP server.\n </Card>\n\n <Card title=\"Eleven v4\" icon=\"duotone waveform\" href=https://elevenlabs.io/"/docs/overview/capabilities/text-to-speech/eleven-v4/">/n What changed in Eleven v4, with voice cloning comparisons.\n </Card>\n</CardGroup>\n\n### Voice quality\n\nFor real-time applications, Flash v2.5 provides ultra-low 75ms latency, while Multilingual v2 delivers the highest quality audio with more nuanced expression.\n\n<CardGroup cols={2} rows={2}>\n <Card title=\"Eleven v4\" href=https://elevenlabs.io/"/docs/overview/models#eleven-v4\">\n Our most emotive, high quality speech synthesis model\n\n <div>\n <div>\n Exceptional voice cloning capabilities\n </div>\n\n <div>\n 90+ languages supported\n </div>\n\n <div>\n 10,000 character limit\n </div>\n\n <div>\n Support for natural multi-speaker dialogue\n </div>\n </div>\n </Card>\n\n <Card title=\"Eleven v4 Turbo\" href=https://elevenlabs.io/"/docs/overview/models#eleven-v4-turbo\">\n Our most emotive, real-time speech synthesis model\n\n <div>\n <div>\n Ultra-low latency (median inference latency of ~100ms†)\n </div>\n\n <div>\n Exceptional voice cloning capabilities\n </div>\n\n <div>\n 90+ languages supported\n </div>\n\n <div>\n Audio tags for fine-grained control\n </div>\n </div>\n </Card>\n\n <Card title=\"Eleven v3\" href=https://elevenlabs.io/"/docs/overview/models#eleven-v3\">\n Our emotionally rich, expressive speech synthesis model\n\n <div>\n <div>\n Dramatic delivery and performance\n </div>\n\n <div>\n 70+ languages supported\n </div>\n\n <div>\n 5,000 character limit\n </div>\n\n <div>\n Support for natural multi-speaker dialogue\n </div>\n </div>\n </Card>\n\n <Card title=\"Eleven v3 Conversational\" href=https://elevenlabs.io/"/docs/overview/models#eleven-v3-conversational\">\n Our expressive, realtime speech synthesis model\n\n <div>\n <div>\n Low latency (~280ms)\n </div>\n\n <div>\n Dramatic delivery and performance\n </div>\n\n <div>\n 70+ languages supported\n </div>\n\n <div>\n Audio tags for fine-grained control\n </div>\n </div>\n </Card>\n\n <Card title=\"Eleven Multilingual v2\" href=https://elevenlabs.io/"/docs/overview/models#multilingual-v2\">\n Lifelike, consistent quality speech synthesis model\n\n <div>\n <div>\n Natural-sounding output\n </div>\n\n <div>\n 29 languages supported\n </div>\n\n <div>\n 10,000 character limit\n </div>\n\n <div>\n Most stable on long-form generations\n </div>\n </div>\n </Card>\n\n <Card title=\"Eleven Flash v2.5\" href=https://elevenlabs.io/"/docs/overview/models#flash-v25\">\n Our fast, affordable speech synthesis model\n\n <div>\n <div>\n Ultra-low latency (~75ms†)\n </div>\n\n <div>\n 32 languages supported\n </div>\n\n <div>\n 40,000 character limit\n </div>\n\n <div>\n Faster model, 50% lower price per character for API generations\n </div>\n </div>\n </Card>\n</CardGroup>\n\n<div>\n <div>\n [Explore all](/docs/overview/models)\n </div>\n</div>\n\n### Voice options\n\nElevenLabs offers thousands of voices across 32 languages through multiple creation methods:\n\n* [Voice library](/docs/overview/capabilities/voices) with 10,000+ community-shared voices\n* [Professional voice cloning](/docs/overview/capabilities/voices#cloned) for highest-fidelity replicas\n* [Instant voice cloning](/docs/overview/capabilities/voices#cloned) for quick voice replication\n* [Voice design](/docs/overview/capabilities/voices#voice-design) to generate custom voices from text descriptions\n\nLearn more about our [voice options](/docs/overview/capabilities/voices).\n\n### Audio effects\n\nAudio effects post-process generated speech so it sounds like it was recorded somewhere specific, without editing it afterwards. Set any combination of:\n\n* **Filter preset**: colors the voice, for example a phone line, an old radio or a robot.\n* **Distance**: moves the speaker away from the microphone.\n* **Environment**: places the voice in a reverberant space, such as a small room or a hall.\n* **Background noise**: loops an ambient bed, such as a café or keyboard typing, under the voice.\n* **Pan**: positions the voice between the left and right channels.\n\nEffects are available on the Text to Speech and Text to Dialogue endpoints, including streaming and timestamped requests, through the [`audio_effects`](/docs/api-reference/text-to-speech/convert#request.body.audio_effects) parameter. List the available filter presets, environments and background noises with the [audio effects catalog](/docs/api-reference/audio-effects/get-catalog). Agents can apply a voice filter too; see [Audio environment](/docs/eleven-agents/customization/voice/audio-environment). In the app, add effects from the [Text to Speech playground](/docs/eleven-creative/playground/text-to-speech).\n\nEffects do not change the cost of a request.\n\n<Warning>\n Reverb (an `environment_id` with a `send_level` above 0) and a non-zero `pan` produce stereo\n audio. Requests that combine them with the mono `pcm_*`, `ulaw_*` or `alaw_*` output formats are\n rejected with a 400 error.\n</Warning>\n\n<Accordion title=\"Supported output formats\">\n The default response format is `mp3`, but other formats like `pcm` and `ulaw` are available.\n\n * **MP3**\n * Sample rates: 22.05kHz - 44.1kHz\n * Bitrates: 32kbps - 192kbps\n * 22.05kHz @ 32kbps\n * 44.1kHz @ 32kbps, 64kbps, 96kbps, 128kbps, 192kbps\n * **PCM (S16LE)**\n * Sample rates: 16kHz - 44.1kHz\n * Bitrates: 8kHz, 16kHz, 22.05kHz, 24kHz, 44.1kHz, 48kHz\n * 16-bit depth\n * **μ-law**\n * 8kHz sample rate\n * Optimized for telephony applications\n * **A-law**\n * 8kHz sample rate\n * Optimized for telephony applications\n * **Opus**\n * Sample rate: 48kHz\n * Bitrates: 32kbps - 192kbps\n\n <Success>\n Higher quality audio options are only available on paid tiers - see our [pricing\n page](https://elevenlabs.io/pricing/api) for details.\n </Success>\n</Accordion>\n\n### Supported languages\n\nOur multilingual v2 models support 29 languages:\n\n*English (USA, UK, Australia, Canada), Japanese, Chinese, German, Hindi, French (France, Canada), Korean, Portuguese (Brazil, Portugal), Italian, Spanish (Spain, Mexico), Indonesian, Dutch, Turkish, Filipino, Polish, Swedish, Bulgarian, Romanian, Arabic (Saudi Arabia, UAE), Czech, Greek, Finnish, Croatian, Malay, Slovak, Danish, Tamil, Ukrainian & Russian.*\n\nFlash v2.5 supports 32 languages - all languages from v2 models plus:\n\n*Hungarian, Norwegian & Vietnamese*\n\nSimply input text in any of our supported languages and select a matching voice from our [voice library](https://elevenlabs.io/app/voice-library). For the most natural results, choose a voice with an accent that matches your target language and region.\n\n### Prompting\n\nThe models interpret emotional context directly from the text input. For example, adding\ndescriptive text like \"she said excitedly\" or using exclamation marks will influence the speech\nemotion. Voice settings like Stability and Similarity help control the consistency, while the\nunderlying emotion comes from textual cues.\n\nRead the [prompting guide](/docs/overview/capabilities/text-to-speech/best-practices) for more details.\n\n<Note>\n Descriptive text will be spoken out by the model and must be manually trimmed or removed from the\n audio if desired.\n</Note>\n\n## FAQ\n\n<AccordionGroup>\n <Accordion title=\"Can I clone my own voice?\">\n Yes, you can create [instant voice clones](/docs/overview/capabilities/voices#cloned) of your own voice\n from short audio clips. For high-fidelity clones, check out our [professional voice\n cloning](/docs/overview/capabilities/voices#cloned) feature.\n </Accordion>\n\n <Accordion title=\"Do I own the audio output?\">\n Yes. You retain ownership of any audio you generate. However, commercial usage rights are only\n available with paid plans. With a paid subscription, you may use generated audio for commercial\n purposes and monetize the outputs if you own the IP rights to the input content.\n </Accordion>\n\n <Accordion title=\"What qualifies as a free regeneration?\">\n A free regeneration allows you to regenerate the same text to speech content without additional cost, subject to these conditions:\n\n * You can regenerate each piece of content up to 2 times for free\n * The content must be exactly the same as the previous generation. Any changes to the text, voice settings, or other parameters will require a new, paid generation\n\n Free regenerations are useful in case there is a slight distortion in the audio output. According to ElevenLabs' internal benchmarks, regenerations will solve roughly half of issues with quality, with remaining issues usually due to poor training data.\n </Accordion>\n\n <Accordion title=\"How do I reduce latency for real-time cases?\">\n Use the low-latency Flash [models](/docs/overview/models) (Flash v2 or v2.5) optimized for near real-time\n conversational or interactive scenarios. See our [latency optimization\n guide](/docs/eleven-api/guides/how-to/best-practices/latency-optimization) for more details.\n </Accordion>\n\n <Accordion title=\"Why is my output sometimes inconsistent?\">\n The models are nondeterministic. For consistency, use the optional [seed\n parameter](/docs/api-reference/text-to-speech/convert#request.body.seed), though subtle\n differences may still occur.\n </Accordion>\n\n <Accordion title=\"What's the best practice for large text conversions?\">\n Split long text into segments and use streaming for real-time playback and efficient processing.\n To maintain natural prosody flow between chunks, include [previous/next text or previous/next\n request id parameters](/docs/api-reference/text-to-speech/convert#request.body.previous_text).\n </Accordion>\n</AccordionGroup>\n\n## Key facts\n\n* **Determinism**: Output is nondeterministic — use the `seed` parameter for more consistent results\n* **Audio effects**: Add a filter preset, distance, reverb, background noise or panning with `audio_effects` at no extra cost\n* **Free regenerations**: Up to 2 free regenerations per generation (same content and parameters only)\n* **Ownership**: You retain ownership of generated audio; commercial use requires a paid plan\n* **Low-latency use cases**: Use Flash models (`eleven_flash_v2` or `eleven_flash_v2_5`) — see the [latency optimization guide](/docs/eleven-api/guides/how-to/best-practices/latency-optimization)\n* **Large text**: Split long text into segments; use the `previous_text` / `next_text` parameters to maintain natural prosody across chunks\n"],"perViewerCopy":[0,true],"askAiEnabled":[0,false],"skillsConfig":[0],"availability":[0],"pageActions":[0,{"options":[0,{"copyPage":[0,true],"viewAsMarkdown":[0,true],"askAi":[0,true],"openAi":[0,true],"claude":[0,true],"cursor":[0,true],"claudeCode":[0,true],"vscode":[0,false],"custom":[1,[[0,{"title":[0,"LLMs full"],"subtitle":[0,"See the entire documentation site as markdown for LLMs"],"url":[0,"https://elevenlabs.io/docs/llms-full.txt"],"icon":[0,"fa-solid fa-file-code"]}]]]}]}],"pageActionsStyle":[0,"default"],"pageActionFiles":[0],"showRssFeedButton":[0,false],"showPageActions":[0,true]}" ssr client="deferred" opts="{"name":"PageHeaderIsland","value":true}" await-children>
Capabilities

Text to Speech

Learn how to turn text into lifelike spoken audio with ElevenLabs.

Overview

ElevenLabs Text to Speech (TTS) API turns text into lifelike audio with nuanced intonation, pacing and emotional awareness. Our models adapt to textual cues across 32 languages and multiple voice styles and can be used to:

  • Narrate global media campaigns & ads
  • Produce audiobooks in multiple languages with complex emotional delivery
  • Stream real-time audio from text

Listen to a sample:

Explore our voice library to find the perfect voice for your project.

The voice library is not available via the API to free tier users.
Products

Step-by-step guide for using text to speech in ElevenLabs.

Developers

Learn how to integrate text to speech into your application.

API reference

Full API reference for the Text to Speech endpoint.

MCP

Generate speech from Claude or any MCP client with the hosted MCP server.

Eleven v4

What changed in Eleven v4, with voice cloning comparisons.

Voice quality

For real-time applications, Flash v2.5 provides ultra-low 75ms latency, while Multilingual v2 delivers the highest quality audio with more nuanced expression.

Eleven v4

Our most emotive, high quality speech synthesis model

Exceptional voice cloning capabilities
90+ languages supported
10,000 character limit
Support for natural multi-speaker dialogue
Eleven v4 Turbo

Our most emotive, real-time speech synthesis model

Ultra-low latency (median inference latency of ~100ms†)
Exceptional voice cloning capabilities
90+ languages supported
Audio tags for fine-grained control
Eleven v3

Our emotionally rich, expressive speech synthesis model

Dramatic delivery and performance
70+ languages supported
5,000 character limit
Support for natural multi-speaker dialogue
Eleven v3 Conversational

Our expressive, realtime speech synthesis model

Low latency (~280ms)
Dramatic delivery and performance
70+ languages supported
Audio tags for fine-grained control
Eleven Multilingual v2

Lifelike, consistent quality speech synthesis model

Natural-sounding output
29 languages supported
10,000 character limit
Most stable on long-form generations
Eleven Flash v2.5

Our fast, affordable speech synthesis model

Ultra-low latency (~75ms†)
32 languages supported
40,000 character limit
Faster model, 50% lower price per character for API generations

Voice options

ElevenLabs offers thousands of voices across 32 languages through multiple creation methods:

Learn more about our voice options.

Audio effects

Audio effects post-process generated speech so it sounds like it was recorded somewhere specific, without editing it afterwards. Set any combination of:

  • Filter preset: colors the voice, for example a phone line, an old radio or a robot.
  • Distance: moves the speaker away from the microphone.
  • Environment: places the voice in a reverberant space, such as a small room or a hall.
  • Background noise: loops an ambient bed, such as a café or keyboard typing, under the voice.
  • Pan: positions the voice between the left and right channels.

Effects are available on the Text to Speech and Text to Dialogue endpoints, including streaming and timestamped requests, through the audio_effects parameter. List the available filter presets, environments and background noises with the audio effects catalog. Agents can apply a voice filter too; see Audio environment. In the app, add effects from the Text to Speech playground.

Effects do not change the cost of a request.

Reverb (an environment_id with a send_level above 0) and a non-zero pan produce stereo audio. Requests that combine them with the mono pcm_*, ulaw_* or alaw_* output formats are rejected with a 400 error.

The default response format is mp3, but other formats like pcm and ulaw are available.

  • MP3
    • Sample rates: 22.05kHz - 44.1kHz
    • Bitrates: 32kbps - 192kbps
    • 22.05kHz @ 32kbps
    • 44.1kHz @ 32kbps, 64kbps, 96kbps, 128kbps, 192kbps
  • PCM (S16LE)
    • Sample rates: 16kHz - 44.1kHz
    • Bitrates: 8kHz, 16kHz, 22.05kHz, 24kHz, 44.1kHz, 48kHz
    • 16-bit depth
  • μ-law
    • 8kHz sample rate
    • Optimized for telephony applications
  • A-law
    • 8kHz sample rate
    • Optimized for telephony applications
  • Opus
    • Sample rate: 48kHz
    • Bitrates: 32kbps - 192kbps

Higher quality audio options are only available on paid tiers - see our pricing page for details.

Supported languages

Our multilingual v2 models support 29 languages:

English (USA, UK, Australia, Canada), Japanese, Chinese, German, Hindi, French (France, Canada), Korean, Portuguese (Brazil, Portugal), Italian, Spanish (Spain, Mexico), Indonesian, Dutch, Turkish, Filipino, Polish, Swedish, Bulgarian, Romanian, Arabic (Saudi Arabia, UAE), Czech, Greek, Finnish, Croatian, Malay, Slovak, Danish, Tamil, Ukrainian & Russian.

Flash v2.5 supports 32 languages - all languages from v2 models plus:

Hungarian, Norwegian & Vietnamese

Simply input text in any of our supported languages and select a matching voice from our voice library. For the most natural results, choose a voice with an accent that matches your target language and region.

Prompting

The models interpret emotional context directly from the text input. For example, adding descriptive text like “she said excitedly” or using exclamation marks will influence the speech emotion. Voice settings like Stability and Similarity help control the consistency, while the underlying emotion comes from textual cues.

Read the prompting guide for more details.

Descriptive text will be spoken out by the model and must be manually trimmed or removed from the audio if desired.

FAQ

Yes, you can create instant voice clones of your own voice from short audio clips. For high-fidelity clones, check out our professional voice cloning feature.

Yes. You retain ownership of any audio you generate. However, commercial usage rights are only available with paid plans. With a paid subscription, you may use generated audio for commercial purposes and monetize the outputs if you own the IP rights to the input content.

A free regeneration allows you to regenerate the same text to speech content without additional cost, subject to these conditions:

  • You can regenerate each piece of content up to 2 times for free
  • The content must be exactly the same as the previous generation. Any changes to the text, voice settings, or other parameters will require a new, paid generation

Free regenerations are useful in case there is a slight distortion in the audio output. According to ElevenLabs’ internal benchmarks, regenerations will solve roughly half of issues with quality, with remaining issues usually due to poor training data.

Use the low-latency Flash models (Flash v2 or v2.5) optimized for near real-time conversational or interactive scenarios. See our latency optimization guide for more details.

The models are nondeterministic. For consistency, use the optional seed parameter, though subtle differences may still occur.

Split long text into segments and use streaming for real-time playback and efficient processing. To maintain natural prosody flow between chunks, include previous/next text or previous/next request id parameters.

Key facts

  • Determinism: Output is nondeterministic — use the seed parameter for more consistent results
  • Audio effects: Add a filter preset, distance, reverb, background noise or panning with audio_effects at no extra cost
  • Free regenerations: Up to 2 free regenerations per generation (same content and parameters only)
  • Ownership: You retain ownership of generated audio; commercial use requires a paid plan
  • Low-latency use cases: Use Flash models (eleven_flash_v2 or eleven_flash_v2_5) — see the latency optimization guide
  • Large text: Split long text into segments; use the previous_text / next_text parameters to maintain natural prosody across chunks