Model catalog
Voice models
Browse every available voice model, organized by family and ready to open with the right creation workflow.
31 models8 families
ElevenLabs
ElevenLabs · 13 models
ElevenLabsNewElevenLabs DubbingDub any video or audio into another language while preserving the original speaker voices using ElevenLabs.
ElevenLabsNewElevenLabs Dubbing v2Translate recorded audio or video while preserving the speaker's emotion, timing and voice with Dubbing v2. Returns dubbed audio. Supports 90+ languages; ArtEmotion accepts clips up to 10 minutes. Film, TV and comparable commercial entertainment uses require separate ElevenLabs authorization.
ElevenLabsNewElevenLabs Eleven v3Expressive Eleven v3 speech with emotional delivery, audio tags and 70+ languages. Use Eleven v4 for the newest expressive model.
ElevenLabsNewElevenLabs Flash v2.5Fast, affordable speech generation with Flash v2.5 across 32 languages. Generate narration and voiceovers with up to 40,000 characters per request.
ElevenLabsNewElevenLabs Instant Voice CloneCreate a reusable ElevenLabs voice from a short, clean recording and hear it speak your preview text immediately.
ElevenLabsNewElevenLabs Multilingual v2ElevenLabs' flagship multilingual TTS model with exceptional voice cloning and expressiveness across 29 languages.
ElevenLabsNewElevenLabs Text to DialogueGenerate multi-speaker dialogue audio by assigning each line of dialogue to an ElevenLabs voice. Powered by Eleven v3.
ElevenLabsNewElevenLabs v3 ConversationalExpressive speech with Eleven v3 Conversational in 70+ languages. ArtEmotion generates one voice per job through the provider's Text to Dialogue WebSocket and saves a playable MP3. Use audio tags such as [laughs] and [whispers]. Maximum 5,000 characters per job.
ElevenLabsNewElevenLabs v4Expressive Eleven v4 narration with improved voice fidelity, audio tags and over 90 languages. Supports premade voices and your authorized voice clones. HTTP jobs are limited to 2,000 characters for reliable generation.
ElevenLabsNewElevenLabs v4 DialogueMulti-speaker dialogue with Eleven v4: up to ten premade voices, expressive audio tags and natural turn-taking. ArtEmotion limits each dialogue job to 2,000 total characters for reliable HTTP generation.
ElevenLabsNewElevenLabs v4 TurboLow-latency expressive Eleven v4 Turbo speech. ArtEmotion renders one voice per prepaid job and saves a complete MP3; this is not a live browser voice session.
ElevenLabsNewElevenLabs Voice ChangerTransform the voice in an audio file to any ElevenLabs pre-made voice while preserving the original speech rhythm and content.
ElevenLabsNewElevenLabs Voice DesignGenerate a brand-new voice from a text description using ElevenLabs Voice Design v3. The model produces an audio sample of the custom voice speaking your sample text. Ideal for prototyping unique narrator voices, character voices, or brand-specific voices.Kokoro
Kokoro · 8 models
KokoroNewKokoro (Brazilian Portuguese)Kokoro lightweight TTS — natural Brazilian Portuguese voices with fast inference.
KokoroNewKokoro (British English)Kokoro lightweight TTS — natural British English voices with fast inference.
KokoroNewKokoro (French)Kokoro lightweight TTS — natural French voices with fast inference.
KokoroNewKokoro (Hindi)Kokoro lightweight TTS — natural Hindi voices with fast inference.
KokoroNewKokoro (Italian)Kokoro lightweight TTS — natural Italian voices with fast inference.
KokoroNewKokoro (Japanese)Kokoro lightweight TTS — natural Japanese voices with fast inference.
KokoroNewKokoro (Spanish)Kokoro lightweight TTS — natural Spanish voices with fast inference.
KokoroNewKokoro (US English)Kokoro lightweight TTS — natural American English voices with fast inference.MiniMax
MiniMax · 3 models
MiniMaxNewMiniMax Speech-02 HDMiniMax's high-definition TTS model with 37-language support and expressive voice personas.
MiniMaxNewMiniMax Speech-02 TurboMiniMax's faster, more affordable TTS model with the same 37-language coverage and voice personas as Speech-02 HD at a lower cost.
MiniMaxNewMiniMax Voice CloneClone any voice from a 10+ second audio sample using MiniMax's voice cloning model.Qwen
Alibaba · 3 models
AlibabaNewQwen 3 TTS Flash (Alibaba)Alibaba's commercial Qwen 3 TTS Flash via DashScope — fast, natural multilingual speech with 17 preset voices and dialect support.
AlibabaNewQwen 3 TTS Voice Design (1.7B)Design a custom voice from a text description and synthesize speech in that voice using Qwen 3 TTS.
AlibabaNewQwen 3 Voice CloneClone a voice and hear it speak in one step — upload a 10+ second recording, write what it should say, and Qwen 3 TTS answers in the cloned voice. The reusable speaker embedding is saved alongside the audio.Gemini
Google · 1 model
Inworld
Inworld · 1 model
Orpheus
Canopy Labs · 1 model
Seed
Seed · 1 model
