Eleven v4
✓ Premade voices ✓ Voice generation ✓ Instant voice cloning ✓ Professional voice cloning Eleven v4 is typically our recommended model for the most natural, human-like narration. It handles context, emphasis, and pacing particularly well, helping speech flow naturally across sentences and paragraphs. It also produces highly accurate voice clones that retain more of the original speaker’s accent, character, and delivery style.- Languages: 87 languages
- Voice settings: Expressiveness, Similarity, Multilingual mode
- Pronunciation rules: Substitute, Say as word, Say as letter sequence
Eleven v4 Turbo
✓ Premade voices ✓ Voice generation ✓ Instant voice cloning ✓ Professional voice cloning Eleven v4 Turbo offers many of the same strengths and limitations as v4 for BeyondWords publishers. If both models are available for your voice, we recommend comparing them in the Editor to see which you prefer.- Languages: 87 languages
- Voice settings: Expressiveness, Similarity, Multilingual mode
- Pronunciation rules: Substitute, Say as word, Say as letter sequence
Eleven v3
✓ Premade voices ✓ Voice generation ✓ Instant voice cloning Eleven v3 is an expressive and dynamic model. It offers particularly strong text normalization, verbalization, and pronunciation across supported languages. However, it does not support request stitching, so you may notice changes in the voice or delivery between paragraphs.- Languages: 74 languages
- Voice settings: Expressiveness, Multilingual mode
- Pronunciation rules: Substitute, Say as word, Say as letter sequence
Eleven v3 Conversational
✓ Premade voices ✓ Voice generation ✓ Instant voice cloning Eleven v3 Conversational offers many of the same strengths and limitations as v3 for BeyondWords publishers. If both models are available for your voice, we recommend comparing them in the Editor to see which you prefer.- Languages: 74 languages
- Voice settings: Expressiveness, Multilingual mode
- Pronunciation rules: Substitute, Say as word, Say as letter sequence
Eleven Multilingual v2
✓ Premade voices ✓ Voice generation ✓ Instant voice cloning ✓ Professional voice cloning Multilingual v2 is a robust and stable option for long-form narration. It supports request stitching, helping the voice and delivery remain consistent between paragraphs. Its pronunciation and text normalization are not as advanced as v3, but it’s a safe choice for smooth, predictable narration.- Languages: 29 languages
- Voice settings: Multilingual mode, Speaking rate, Expressiveness, Similarity, Style, Speaker boost
- Pronunciation rules: Substitute, Say as word, Say as letter sequence
Multilingual v2 is not the only multilingual model. All voices can speak the languages supported by their model, though quality may vary.
Eleven Flash v2.5
✓ Premade voices ✓ Voice generation ✓ Instant voice cloning ✓ Professional voice cloning Flash v2.5 is a fast, stable model that maintains consistent delivery between paragraphs.- Languages: 32 languages
- Voice settings: Multilingual mode, Speaking rate, Expressiveness, Similarity
- Pronunciation rules: Substitute, Say as word, Say as letter sequence, Phonetic spelling (English only)
FAQs
How do I set the voice model via the API?
How do I set the voice model via the API?
If you’re using the API, each voice and model combination has a unique
voice_id, so you don’t need to set the model separately.How does language support work across voice models?
How does language support work across voice models?
Each voice model supports a specific set of languages. Any voice can speak any language supported by the selected model, although voices typically perform best in their native language and accent.
How do I choose the right voice model?
How do I choose the right voice model?
v4 is typically the best choice for natural, human-like narration. If more than one model is available for your voice, listen to a short preview when making your selection. For a longer, more representative sample, create example content in the Editor.
Is the newer model always better?
Is the newer model always better?
Not necessarily. Newer models may offer improvements in some areas, but the best choice can depend on your language, voice type, and the kind of content you’re generating.Compare the available options in the Editor if you want to test which sounds best for your content.