IA ñe'ẽasa ñehendurã
Oñemoambue jehaipy ñe'ẽte he'íva natural-pe modelo IA fuente abierto rupive. Ojeporuve'ỹre, ndaha'éi oñeikotevẽva peteĩ cuenta.
Ojehaijey ñe'ẽnguéra etiquetas SSML-pe peteĩ control hekopete g̃uarã:
<speak><prosody rate="slow">Slow speech</prosody></speak>
Omoĩnge umi marcador de emoción ombyai hag̃ua entrega (modelo apo iñambue):
Oñemohenda ñe'ẽnguéra ojehechapyréva (tembiapo = ñe'ẽnguéra):
Modelo detalle-kuéra
Kokoro
Kokoro is an 82 million parameter text-to-speech model that punches well above its weight class. Despite its tiny size, it produces remarkably natural and expressive speech. Kokoro supports multiple languages including English, Japanese, Chinese, and Korean with a variety of expressive voices. It runs incredibly fast — generating audio nearly 100x faster than real-time on a GPU.
| Desarrollador: | Hexgrad |
| Licencia: | Apache 2.0 |
| Velocidad | Fast |
| Calidad: | |
| Ñe'ẽ | 8 Ñe'ẽ |
| VRAM | 1.5GB |
| Clonación ñe'ẽnguéra rehe | No admitido |
Ñemomarandu ojehupyty hag̃ua peteĩ mba'e porãve
- Oiporu puntuación oĩporãva pausa ha entonación natural-pe g̃uarã.
- Oñe'ẽ papapy ha abreviatura-kuéra ñe'ẽporãve hag̃ua
- Omoĩngeve comillas omoheñói hag̃ua pausa michĩva ñe'ẽjoaju apytépe
- Oiporu punto de suspensión (...) pausa dramática pukukue
- Oñeha'ã Kokoro térã CosyVoice 2 ojehupyty hag̃ua resultado natural
- Oiporu Dia diálogo-pe heta ñehendurã ha podcast-kuéra rehegua contenido
Caracter jeporu
| Ta'ãnga | Presupuesto peteĩteĩ 1K carácter-pe g̃uarã |
|---|---|
| Libre | 1:1 (tembiapo) |
| Estándar | Caracteres 2x |
| Premium | 4x caractere |
Mba'éichapa IA rembiapo ñe'ẽjoaju rehegua
Oñemoheñói ñe'ẽnguéra peteĩteĩ mbohapy paso ndahasýiva rupive. Nahániri conocimiento técnico.
Oike'ỹre ñe'ẽ
Ehai, ape'a térã oguerahauka jey ñe'ẽ ojehe'a hag̃ua ñe'ẽ'apo. Omoneĩ 5000 caractere peve peteĩ generación-pe g̃uarã umi cuenta libre-pe g̃uarã, térã 100000 umi plan ojejapyhyhápe. Oiporu texto ndaha'éiva térã omoĩnge etiquetas SSML control avanzado ñe'ẽ'apo, pausa ha enfasis rehegua.
Oñeporavo modelo ha ñe'ẽ
Oñeporavo 20+ modelo IA apytépe mbohapy nivel-pe. Oñeporavo peteĩ ñe'ẽ ojokupytýva nde contenido ndive, oporavo ñe'ẽ ojehupytyséva, omohenda reproducción rehegua velocidad 0.5x guive 2.0x peve ha oporavo formato de salida ojehecharamovéva (MP3, WAV, OGG térã FLAC).
Generar y descargar
Ohesa'ỹijo peteĩ enlace compartible. Oiporu API proceso por lotes ha ñemoĩnge hag̃ua tembiapo jehaipy ryepýpe.
Ojeporukuaáva ñe'ẽjoaju ñeikumbyrã
IA rupive ojehaíva ñe'ẽngue-pe omoambue mba'éichapa yvypóra omoheñói, oiporu ha ojokupyty umi contenido sonoro heta industria-pe.
Opaite modelo ñe'ẽ'apo rehegua
Especificación detallada peteĩteĩ modelo IA oĩva TTS.ai-pe. Oñembojoja calidad, velocidad, ñe'ẽjoaju ha característica-kuéra ojejuhu hag̃ua modelo iporãvéva nde proyecto-pe g̃uarã.
Kokoro
Free
Kokoro hína peteĩ modelo texto-gui-ñe'ẽ-pe g̃uarã oguerekóva 82 millón parámetro, ha'éva tuichaiterei mba'e ijyvatevéva clase de peso-gui. Jepéramo michĩeterei, ome'ẽ ñe'ẽnguéra peteĩteĩ natural ha expresiva. Kokoro oykeko heta ñe'ẽnguérape, oikehápe inglés, japonés, chino ha coreano heta ñe'ẽnguéra expresiva-kuéra. Ojeporu pya'eterei, omoheñói ñe'ẽnguéra 100-guive pya'eve peteĩ GPU tiempo real-pe.
Hexgrad
Apache 2.0
Fast
en, ja, zh, fr, it, pt, es, hi
1.5GB
No
Libre
Piper
Free
Piper hína peteĩ motor de texto-gui-ñe'ẽ-pe pyahu, omoheñóiva Rhasspy oiporúvo VITS ha larynx arquitectura. Ojeporuporã CPU-pe, ha upéva ombohekoporãve umi dispositivo periférico-pe, automatización hogapypegua ha aplicación oikotevẽva TTS fuera de línea. Oguerekóvo hetave 100 ñe'ẽ 30 ñe'ẽgui, Piper ome'ẽ ñe'ẽ ñeikumby natural tiempo real-pe, avei Raspberry Pi 4-pe.
Rhasspy
MIT
Fast
en, de, fr, es, it, pt, nl, pl, ru, zh, ar, cs, da, fi, el, hu, is, ka, kk, ne, no, ro, sk, sr, sv, sw, tr, uk, vi, ca, cy, fa, lv, sl, lb
0 (CPU only)
No
Libre
VITS
Free
VITS (inferencia variacional aprendizaje adversario rupive ñe'ẽ'aravo ñe'ẽ'aravo ñe'ẽ'aravo-gui ñe'ẽ'aravo-pe g̃uarã) ha'e peteĩ método TTS paralelo, oguerekóva peteĩ ñe'ẽ'aravo natural umi modelo ko'agãgua mokõi etapa-guive, oipuru inferencia variacional oñembohetavehápe flujo normalización ha peteĩ proceso de aprendizaje adversario, ha upéicha ojehupyty peteĩ naturalidad oñemoporãvehápe.
Jaehyeon Kim et al.
MIT
Fast
en, de, es, fr, pt, nl, fi, hu, bg, ja, pl
1GB
No
Libre
MeloTTS
Free
MeloTTS MyShell.ai mba'éva hína peteĩ biblioteca TTS multilenguaje rehegua oykekóva inglés (americano, británico, indio, australiano), español, francés, chino, japonés ha coreano. Ipya'eeterei, omboheko umi jehaipy pya'eterei rupi CPU añónte. MeloTTS oñemohenda ojeporu hag̃ua producción-pe ha oykeko CPU ha GPU inferencia.
MyShell.ai
MIT
Fast
en, es, fr, zh, ja, ko
0.5GB (GPU optional)
No
Libre
Bark
Standard
Bark, Suno mba'éva, ha'e peteĩ modelo texto-gui ñe'ẽnguérape g̃uarã oñemopyendáva transformador-pe, ikatúva omoheñói ñe'ẽnguéra peteĩteĩ ha'evéva, ha ambue ñe'ẽnguéra, taha'e purahéi, ru he'ẽ ha efecto sonoro. Ikatu omoheñói ñe'ẽnguéra ndaha'éiva ñe'ẽnguéra, taha'e ñe'ẽnguéra ñemboyke, ñe'ẽnguéra ñemboyke ha ñe'ẽnguéra ñemboyke. Bark oykeko hetave 100 preconfiguración ñe'ẽnguérape g̃uarã ha hetave 13 ñe'ẽ.
Suno
MIT
Slow
en, zh, fr, de, hi, it, ja, ko, pl, pt, ru, es, tr
5GB
No
2x
Bark Small
Standard
Bark Small hína peteĩ versión destilada modelo Bark mba'éva, ombohasahápe peteĩ ñe'ẽnguéra calidad inferencia rehegua velocidad pya'eve ha memoria rehegua requisito michĩvéva rehe, ha oguereko gueteri Bark capacidad omongu'éva ñe'ẽnguéra emoción, ta'ãnga ha heta ñe'ẽ rupive.
Suno
MIT
Medium
en, zh, fr, de, hi, it, ja, ko, pl, pt, ru, es, tr
2GB
No
2x
CosyVoice 2
Standard
CosyVoice 2, Tongyi Lab Alibaba mba'éva, ohupyty ñe'ẽnguéra calidad ojokupytýva yvypóra ñe'ẽme, latencia sa'ivéva, ha upéva omboheko ojeporu hag̃ua tiempo real-pe. Oiporu peteĩ enfoque de cuantización escalar finito síntesis de flujo-pe g̃uarã ha oykeko ñe'ẽnguéra clonación cero disparo rehegua, síntesis interlingüística ha control de emoción de grano fino. Oiko porãve heta sistema comercial TTS-gui evaluación subjetiva-pe.
Alibaba (Tongyi Lab)
Apache 2.0
Medium
en, zh, ja, ko, fr, de, it, es
4GB
Ha'e
2x
Dia TTS
Standard
Dia, Nari Labs mba'éva, ha'e peteĩ modelo texto-gui-ñe'ẽ-pe g̃uarã parámetro 1.6B, oñemoheñói va'ekue omoheñói hag̃ua ñe'ẽasa heta ñe'ẽsarekóva. Ikatu omboguata ñe'ẽsasõ oguerekóva ñe'ẽjoaju natural mokõi ñe'ẽsarekóva turno-kuéra, prosodi ha expresión emocional oguerekóva. Dia iñambueporã ojejapo hag̃ua contenido podcast-peguápe, ñe'ẽsasõ aranduka ñe'ẽ'arúpe ha IA ñe'ẽsasõ rehegua.
Nari Labs
Apache 2.0
Medium
en
4GB
No
2x
Parler TTS
Standard
Parler TTS hína peteĩ modelo ñe'ẽ-gui-ñe'ẽ-pe g̃uarã oipuruhápe ñe'ẽnguéra rehegua descripción ñe'ẽnguéra rehe oñeñangareko hag̃ua ñe'ẽnguéra oguerekóva rehe. Oñeha'ã'ỹre peteĩ ñe'ẽnguéra rehe, ojehechauka ñe'ẽnguéra oikotevẽva (techapyrã, "peteĩ kuña hova'ỹva acento británico michĩvéva ndive, oñe'ẽva pya'e ha hesakã'ỹre") ha Parler omoheñói ñe'ẽnguéra ojojoguáva upe descripción rehe. Kóva ojapo ichugui peteĩ flexibilidad ha'evéva umi aplicación creativa-pe g̃uarã.
Hugging Face
Apache 2.0
Medium
en
4GB
No
2x
IndexTTS-2
Standard
IndexTTS-2 hína peteĩ sistema avanzado ñe'ẽ-gui ñe'ẽngue-pe g̃uarã, ojehecharamova'ekue ñe'ẽnguéra síntesis-pe, control de emoción granular rupive, ikatu omoheñói ñe'ẽnguéra peteĩ tono emocional específico rehe, taha'e ñemboyke, py'aguapy, ira térã ñemboyke, oikotevẽ'ỹre umi dato de entrenamiento específico umi emoción rehegua. Ko modelo oipuru umi vector de emoción ombohape hag̃ua ñe'ẽnguéra expresión emocional generada.
Index Team
Bilibili Model License
Medium
en, zh
4GB
Ha'e
2x
Spark TTS
Standard
Spark TTS SparkAudio mba'éva hína peteĩ modelo ñe'ẽ-gui-ñe'ẽ-pe g̃uarã ombojoajuhápe ñe'ẽ clonación emoción ha ñe'ẽnguéra controlable rehe. Oiporukuévo 5 segundo añónte ñe'ẽnguéra referencia-pegua, ikatu omoheñói peteĩ clona peteĩ ñe'ẽme ha upéi omoheñói ñe'ẽnguéra oguerekóva heta emoción, velocidad ha estilo, ha upéicha avei omombaretekuévo ñe'ẽnguéra clonada-kuéra identidad. Spark TTS oipuru peteĩ sistema de control oñemopyendáva prompt-pe.
SparkAudio
CC BY-NC-SA 4.0
Medium
en, zh
4GB
Ha'e
2x
GPT-SoVITS
Standard
GPT-SoVITS ombojoaju lenguaje modelo GPT estilo SoVITS ndive (Inferencia de voz ñe'ẽ ñeikumby ha síntesis rupive) peteĩ clonación de voz potente-rã sa'i toma-pe g̃uarã. 5 segundo sa'ivéva ñe'ẽnguéra referencia-gui, ikatu omoheñói peteĩ clona de voz ha omoheñói ñe'ẽnguéra pyahu, ha upéicha avei oñangareko umi característica ojekuaáva ñe'ẽnguéra rehe. Oĩ porãve ñe'ẽnguéra ha síntesis ñe'ẽnguéra rehe.
RVC-Boss
MIT
Slow
en, zh, ja, ko
6GB
Ha'e
2x
Orpheus
Standard
Orpheus hína peteĩ modelo ñe'ẽnguérape g̃uarã tuichaháicha, ohupytyséva yvypóra ñe'ẽnguéra rehegua expresión emocional; oñemoarandu rire hetave 100.000 hora-pe hetaichagua ñe'ẽnguéra rehegua, ojehecharamo ñe'ẽnguéra rehe ñemboguata, umi temimo'ã natural, ñe'ẽnguéra estilo ha ñe'ẽnguéra rehe ñemboguatápe; Orpheus ikatu omoheñói ñe'ẽnguéra ndojoavyiva'ekue yvypóra ñe'ẽnguéra rehe.
Canopy Labs
Llama 3.2 Community
Medium
en
4GB
No
2x
Chatterbox
Premium
Chatterbox, Resemble AI mba'éva, ha'e hína peteĩ modelo avanzado ñe'ẽ clonación rehegua cero-shot. Ikatu ombohasa oimeraẽ ñe'ẽ peteĩ muestra de sonido-gui peteĩ precisión tuichavéva rehe, ojapyhykuévo ndaha'éi timbre añónte, avei ñe'ẽnguéra estilo ha umi matiz emocional. Chatterbox avei oguereko control emocional iñambuéva, ombohapéva oñemohenda hag̃ua tono emocional ñe'ẽnguéra generada-pegua, ojehecha'ỹre identidad de voz-gui.
Resemble AI
MIT
Medium
en
4GB
Ha'e
4x
Tortoise TTS
Premium
Tortoise TTS hína peteĩ sistema texto-gui ñe'ẽngue-pe g̃uarã, oguerekóva ñe'ẽnguéra ñemoambue autorregresivo, omopyendáva ñe'ẽnguéra rehegua calidad ha'eveve hag̃ua ipya'eve. Oiporu arquitectura DALL-E-pe oñemopyendáva omoheñói hag̃ua ñe'ẽnguéra peteĩteĩ, oguerekóva prosodi ha ñe'ẽnguéra ñembojoja porãva. Jepéramo ipya'eve hína heta alternativa rovake, Tortoise omoheñói ñe'ẽnguéra ñemoambue rehegua modelo realista-véva oĩva peteĩ ecosistema de código abierto-pe.
James Betker
Apache 2.0
Slow
en
8GB
Ha'e
4x
StyleTTS 2
Premium
StyleTTS 2-pe ojehupyty TTS síntesis yvypóra nivel-pegua, ombojoajukuévo estilo difusión ha ñe'ẽnguéra rehe ñembokatupyry oipurukuévo lenguaje modelo tuichaitereíva. Ojapo ñe'ẽnguéra oguerekóva sonido naturalvéva peteĩ ñe'ẽnguéra ñemohenda apytépe, ojoguahápe umi yvypóra ñe'ẽnguéra rehe. StyleTTS 2 oipuru modelo estilo-kuéra oñemohendáva difusión-pe ojapyhy hag̃ua yvypóra ñe'ẽnguéra ñemoambue tuichakue.
Columbia University
MIT
Medium
en
4GB
No
4x
OpenVoice
Premium
OpenVoice MyShell.ai rupive ombohape ñe'ẽnguéra clonación inmediata peteĩ control granular rupive ñe'ẽnguéra estilo, emoción, acento, ritmo, pausa ha entonación rehegua. Ikatu clonar peteĩ ñe'ẽ peteĩ clip de sonido mbyky guive ha omoheñói ñe'ẽnguéra heta ñe'ẽme, ojejavykuévo ñe'ẽnguéra identidad. OpenVoice avei omba'apo ñe'ẽnguéra conversor ramo, ombohapéva ñe'ẽnguéra ñemoambue tiempo real-pe.
MyShell.ai / MIT
MIT
Medium
en, zh, ja, ko, fr, es
4GB
Ha'e
4x
Qwen3 TTS
Standard
Qwen3-TTS hína peteĩ modelo texto-gui ñe'ẽngue-pe g̃uarã oguerekóva 1.700 millón parámetro, Qwen aty Alibaba mba'éva. Oipytyvõ mokõi modo-pe: ñe'ẽnguéra oñemohendapyréva control emocional rupive (9 ñe'ẽnguéra) ha peteĩ modo ñe'ẽnguéra diseño rehegua ojehechaukahápe ñe'ẽnguéra oikotevẽva lenguaje natural-pe. Oipytyvõ 10 ñe'ẽnguérape, oguerekóva expresividad ha prosodi natural ijyvatevéva.
Alibaba (Qwen)
Apache 2.0
Medium
en, zh, ja, ko, de, fr, ru, pt, es, it
7GB
No
2x
VieNeu-TTS-v2
Standard
VieNeu-TTS-v2 hína peteĩ modelo TTS vietnamita ypy oguerekóva 300M parámetro oñemoaranduhápe 10.000 arýrupi dato bilingüe rehe. Oipytyvõkuaa código-remu en-vi, 7 ñe'ẽ oñemohendapyréva oguerekóva acento norte ha sur gotyo, ha ñe'ẽ clonación instantánea 3-5 segundo guive ñe'ẽ referencia rehegua. Oñemongu'e CPU-pe inferencia GGUF Q4 rupive + descodificador ñe'ẽ ONNX — ndaipóri GPU oñeikotevẽva, generación oñemohu'ã 7 segundo rupi. Oñemopu'ã peteĩ backbone Qwen3-pe.
Phạm Nguyễn Ngọc Bảo
Apache 2.0
Fast
vi, en
CPU
Ha'e
2x
Sesame CSM
Premium
Sesame CSM (Modelo de Conversación de Habla) ha'e peteĩ modelo oguerekóva mil millones de parámetro oñemohendáva oñemoheñói hag̃ua ñe'ẽñe'ẽ, omoha'ãnga umi patrón natural yvypóra ñe'ẽñe'ẽ rehegua, oikehápe avei tiempo de turno, ñembohovái canal-pe, reacción emocional ha ñe'ẽñe'ẽ rehegua flujo. CSM omoheñói peteĩ ñe'ẽñe'ẽ he'iséva peteĩ ñe'ẽñe'ẽ yvypóra rehegua, ñe'ẽñe'ẽ sintético rãngue.
Sesame
Apache 2.0
Slow
en
8GB
No
4x
Chatterbox Turbo
Standard
Chatterbox Turbo, Resemble AI mba'éva, ha'e hína peteĩ parámetro 350M-pegua actualización Chatterbox-pe g̃uarã, ome'ẽva 6x-peve velocidad tiempo real-pe peteĩ latencia sa'ive 200ms-gui. Omoneĩ etiqueta paralingüística taha'e [risa], [to'o] ha [chuckle] texto ryepýpe. Oike avei Perth marca de agua opaite umi sonido generado-pe ojejuhu hag̃ua oúva moõguipa.
Resemble AI
MIT
Fast
en
2GB
Ha'e
2x
VoxCPM
Standard
VoxCPM 1.5 OpenBMB mba'éva hína peteĩ modelo TTS pyahu tokenizador-ỹva omba'apóva espacio continuo-pe token discreto-kuéra rangue. Ojapo ñe'ẽnguéra 44.1kHz fidelidad-pe, oykeko ñe'ẽnguéra clonación cero-disparo rehegua 3-10 segundo-pe, ha oguereko consistencia párrafo-kuéra apytépe. Clonación ñe'ẽnguéra ñembojoaju rupive ikatu ojeiporu ñe'ẽnguéra inglés-gui chino-pe g̃uarã ha viceversa.
OpenBMB
Apache 2.0
Fast
en, zh
4GB
Ha'e
2x
Kani TTS 2
Free
Kani-TTS-2 NineNineSix mba'éva hína peteĩ modelo ultraligero 400M parámetro-kuéra rehegua oñemopu'ãva peteĩ backbone LFM2 AI líquido-pe NVIDIA NanoCodec ndive. Ojeporu 3GB VRAM añónte ha ome'ẽ ~10 segundo ñe'ẽnguéra ~2 segundo aja peteĩ A100 (RTF 0.2)-pe. Ko'agãgua versión pública ome'ẽ peteĩ punto de control `kani-tts-2-en` inglés-pe añónte ha ndoguerekói pe gancho de incrustación oñeikotevẽva ñe'ẽnguéra clonación-pe g̃uarã — ojeporu Chatterbox / IndexTTS2 / F5-TTS clonación-pe g̃uarã, térã Kokoro / MeloTTS ndaha'éiva inglés-pe g̃uarã.
NineNineSix
Apache 2.0
Fast
en
3GB
No
Libre
OuteTTS
Free
OuteTTS ombotuichave umi lenguaje modelo tuichavéva oguerekóva capacidad ñe'ẽ'apo-gui ñe'ẽ'apo-pe g̃uarã, ha upéicha avei oñangareko arquitectura ypy rehe. Oipytyvõ heta backend-pe, oikehápe llama.cpp (CPU/GPU), Hugging Face Transformers, ExLlamaV2, VLLM, ha avei inferencia navegador rupive Transformers.js rupive. Oguerekóva clonación ñe'ẽ'apo rehegua cero disparo rupive umi perfil ñe'ẽ'apohára rehegua oñeñongatuhápe JSON-ramo.
OuteAI
Apache 2.0
Fast
en
2GB
Ha'e
Libre
VibeVoice
Standard
VibeVoice Microsoft mba'éva omoheñói ñe'ẽjoaju ipukúva 90 minuto peve, oykekohápe 4 ñe'ẽha'ãnga simultáneo, ha upéva oikoporãve podcast ha ñe'ẽjovake. Realtime 0.5B variante oguereko peteĩ latencia de ~300ms jeporu interactivo-pe g̃uarã. Oipytyvõ ñe'ẽjovake etiqueta-kuérape oñemoheñói hag̃ua ñe'ẽjovake heta turno-pe.
Microsoft
MIT
Fast
en, zh
4GB
No
2x
Pocket TTS
Free
Pocket TTS Kyutai mba'éva (Moshi omoheñói va'ekue) ha'e peteĩ modelo compacto texto-gui ñe'ẽngue-pe g̃uarã oguerekóva 100M parámetro, imbaretevéva ipukuvévagui. Oñemongu'e porã CPU-pe, oykeko ñe'ẽ clonación cero tiro rehegua peteĩ muestra de sonido añónte guive ha omoheñói ñe'ẽngue oguerekóva sonido natural. Modelo michĩva ojapo ichugui peteĩ modelo ideal umi entorno de despliegue extremo ha sa'i recurso oguerekóvape g̃uarã.
Kyutai
MIT
Fast
en, fr
1GB
Ha'e
Libre
Kitten TTS
Free
Kitten TTS KittenML mba'éva hína peteĩ modelo texto-gui-ñe'ẽ-pe g̃uarã ipya'evéva, oñemopyendáva ONNX-pe. Oguerekóva variante 15M guive 80M peve (25-80 MB disco-pe), ome'ẽ ñe'ẽ ñeikumby porã CPU-pe oikotevẽ'ỹre peteĩ GPU. Oguerekóva 8 ñe'ẽ oñemohendáva, ñe'ẽ ñeikumby pya'e oñemoambuéva ha ñe'ẽ ñeikumby preprocesamiento oñemohendáva papapykuéra, moneda ha unidad-kuéra. Iñambueporã umi aplicación desplegable ha de baja latencia-pe g̃uarã.
KittenML
Apache 2.0
Fast
en
0GB
No
Libre
CosyVoice3
Standard
CosyVoice3 hína evolución pyahu FunAudioLLM aty Alibaba mba'éva. Oguerekóva inferencia bi-streaming ~150ms de latencia rehe, control oñemopyendáva instrucción-pe emoción/velocidad/volumen ha ñe'ẽnguéra jojoguaporãve clonación cero-disparo rehegua. Oipytyvõ 9 ñe'ẽ ha 18 dialecto chino. Variante ajustada RL-pe ome'ẽ peteĩ prosodia moderno.
Alibaba (FunAudioLLM)
Apache 2.0
Fast
en, zh, ja, ko, de, es, fr, it, ru
4GB
Ha'e
2x
NAMAA Saudi TTS
Standard
NAMAA Saudi TTS hína peteĩ árabe saudita ñe'ẽnguéra ñemoambue, ChatterboxMultilingual IA-pegua. Oñemoarandu NAMAA Space rupive ñe'ẽnguéra árabe saudita autentica-pe, omoheñói peteĩ árabe estándar moderno ha ñe'ẽnguéra árabe saudita ñe'ẽnguéra ñemoambue, umi modelo multilingüe genérico ndojokupytýi va'ekue. Oguereko Chatterbox ñe'ẽnguéra clonación cero-shot ha control emocional ñe'ẽnguéra referencia-pegua rupive. TTS árabe ypy oguerekóva peso abierto ojeporu TTS.ai-pe.
NAMAA Space
MIT
Medium
ar
6GB
Ha'e
2x
Darwin TTS
Standard
Darwin-TTS-1.7B-Cross FINAL-Bench mba'éva hína peteĩ variante jeporekarã Qwen3-TTS-1.7B rehegua, oĩhápe 84 tensor-FFN ñe'ẽ'apohára (8,6%) ojoajuhápe α=3% rehe tensor ojokupytýva Qwen3-1.7B-Base-gui. Ko combinación oñemohenda ñembokatupyry'ỹ rehe ha ome'ẽ peteĩ clonación ñe'ẽjoaju ojoavýva ojekuaaporãve hag̃ua coreano, inglés, japonés ha chino-pe. Oiko modo de clonación ñe'ẽjoaju cero disparo-pe (3 segundos ñe'ẽjoaju referencia).
FINAL-Bench
Apache 2.0
Medium
en, ko, ja, zh
7GB
Ha'e
2x
MOSS-TTSD
Standard
MOSS-TTSD v1.0 OpenMOSS-pegua hína peteĩ modelo ñe'ẽasa ñe'ẽ'arã 7B-pegua, omboguatahápe ñe'ẽasa peteĩ ñe'ẽ'arã mbykymi guive. Oipytyvõ 5 ñe'ẽha'ãnga simultáneo rupive etiqueta [S1]/[S2], ñe'ẽ clonación cero-disparo rehegua 3-10 s ñe'ẽ'arã referencia rehegua, ha 60 mbyte peve ñe'ẽ'arã coherente multi-giro rehegua 20 ñe'ẽ rupive. Ojoavy MOSS-TTS-gui — TTSD oñemopyenda umi flujo de trabajo podcast/libro ñe'ẽ'arã/doblado-pe g̃uarã.
OpenMOSS
Apache 2.0
Medium
en, zh
12GB
Ha'e
2x
Ming-Omni TTS
Free
Ming-omni-tts-0.5B inclusionAI mba'éva hína peteĩ modelo de discurso omnimodal compacto oñemopyendáva BailingMM columna vertebral densa-pe, oguerekóva peteĩ descodificador de sonido ojokupytýva flujo rehe parche-pa-parche rupive. Oikuave'ẽ peteĩ salida 44.1kHz (CD calidad ykére), oykeko ñe'ẽ clonación cero disparo rehegua peteĩ referencia 3+ segundo rehegua, ha oike emoción/dialecto/BGM control integrado umi instrucción JSON rupive. Estabilidad mbarete - 0.83% WER umi referencia chino-pe.
inclusionAI
Apache 2.0
Medium
en, zh
3GB
Ha'e
Libre
MOSS-TTS Nano
Free
MOSS-TTS-Nano-100M hína peteĩ variante 100M-parámetro-pegua OpenMOSS-pegua, MOSS-TTS familia-pegua, ombojoajuhápe arquitectura de transformador de retraso. Oñemoambue modelo 8B calidad ypy rehe peso michĩvévape ~80 ha VRAM michĩvévape peteĩ ñeikotevẽ rupive, ha upéva ojapose hag̃ua ikatu hína despliegue de nivel libre ha alto rendimiento.
OpenMOSS
Apache 2.0
Fast
en, zh, de, es, fr, ja, it, ko, ru, ar, pt
2GB
Ha'e
Libre
Kokoro
Libre
Kokoro is an 82 million parameter text-to-speech model that punches well above its weight class. Despite its tiny size, it produces remarkably natural and expressive speech. Kokoro supports multiple languages including English, Japanese, Chinese, and Korean with a variety of expressive voices. It runs incredibly fast — generating audio nearly 100x faster than real-time on a GPU.
Hexgrad
Apache 2.0
Fast
Piper
Libre
Piper is a lightweight text-to-speech engine developed by Rhasspy that uses VITS and larynx architectures. It runs entirely on CPU, making it ideal for edge devices, home automation, and applications requiring offline TTS. With over 100 voices across 30+ languages, Piper delivers natural-sounding speech at real-time speeds even on a Raspberry Pi 4.
Rhasspy
MIT
Fast
VITS
Libre
VITS (Variational Inference with adversarial learning for end-to-end Text-to-Speech) is a parallel end-to-end TTS method that generates more natural sounding audio than current two-stage models. It adopts variational inference augmented with normalizing flows and an adversarial training process, achieving a significant improvement in naturalness.
Jaehyeon Kim et al.
MIT
Fast
MeloTTS
Libre
MeloTTS by MyShell.ai is a multilingual TTS library supporting English (American, British, Indian, Australian), Spanish, French, Chinese, Japanese, and Korean. It is extremely fast, processing text at near real-time speed on CPU alone. MeloTTS is designed for production use and supports both CPU and GPU inference.
MyShell.ai
MIT
Fast
Kani TTS 2
Libre
Kani-TTS-2 by NineNineSix is an ultra-lightweight 400M parameter model built on a Liquid AI LFM2 backbone with NVIDIA NanoCodec. It runs in just 3GB VRAM and produces ~10 seconds of speech in ~2 seconds on an A100 (RTF 0.2). The current public release ships an English-only `kani-tts-2-en` checkpoint and does not expose the speaker-embedding hook needed for voice cloning — use Chatterbox / IndexTTS2 / F5-TTS for cloning, or Kokoro / MeloTTS for non-English.
NineNineSix
Apache 2.0
Fast
OuteTTS
Libre
OuteTTS extends large language models with text-to-speech capabilities while preserving the original architecture. It supports multiple backends including llama.cpp (CPU/GPU), Hugging Face Transformers, ExLlamaV2, VLLM, and even browser inference via Transformers.js. Features zero-shot voice cloning through speaker profiles saved as JSON.
OuteAI
Apache 2.0
Fast
Pocket TTS
Libre
Pocket TTS by Kyutai (creators of Moshi) is a compact 100M parameter text-to-speech model that punches well above its weight. It runs efficiently on CPU, supports zero-shot voice cloning from a single audio sample, and produces natural-sounding speech. The small model size makes it ideal for edge deployment and low-resource environments.
Kyutai
MIT
Fast
Kitten TTS
Libre
Kitten TTS by KittenML is an ultra-lightweight text-to-speech model built on ONNX. With variants from 15M to 80M parameters (25-80 MB on disk), it delivers high-quality voice synthesis on CPU without requiring a GPU. Features 8 built-in voices, adjustable speech speed, and built-in text preprocessing for numbers, currencies, and units. Ideal for edge deployment and low-latency applications.
KittenML
Apache 2.0
Fast
Ming-Omni TTS
Libre
Ming-omni-tts-0.5B by inclusionAI is a compact omni-modal speech model built on the BailingMM dense backbone with a Patch-by-Patch flow-matching audio decoder. Delivers 44.1kHz output (near CD quality), supports zero-shot voice cloning from a 3+ second reference, and includes built-in emotion / dialect / BGM control via JSON instructions. Excellent stability — 0.83% WER on Chinese benchmarks.
inclusionAI
Apache 2.0
Medium
MOSS-TTS Nano
Libre
MOSS-TTS-Nano-100M is OpenMOSS's compact 100M-parameter variant of the MOSS-TTS family, sharing the delay-transformer architecture. Trades the 8B model's peak quality for ~80x smaller weights and dramatically lower per-request VRAM, making it suitable for free-tier and high-throughput deployments. Same 20-language reach.
OpenMOSS
Apache 2.0
Fast
Bark
Estándar
Bark by Suno is a transformer-based text-to-audio model that can generate highly realistic, multilingual speech as well as other audio like music, background noise, and sound effects. It can produce nonverbal communications like laughing, sighing, and crying. Bark supports over 100 speaker presets and 13+ languages.
Suno
MIT
Slow
en, zh, fr, de, hi, it, ja, ko, pl, pt, ru, es, tr
No
Bark Small
Estándar
Bark Small is a distilled version of the Bark model that trades some audio quality for significantly faster inference speeds and lower memory requirements. It retains Bark's ability to generate speech with emotions, laughter, and multiple languages.
Suno
MIT
Medium
en, zh, fr, de, hi, it, ja, ko, pl, pt, ru, es, tr
No
CosyVoice 2
Estándar
CosyVoice 2 by Alibaba's Tongyi Lab achieves human-comparable speech quality with extremely low latency, making it ideal for real-time applications. It uses a finite scalar quantization approach for streaming synthesis and supports zero-shot voice cloning, cross-lingual synthesis, and fine-grained emotion control. It outperforms many commercial TTS systems in subjective evaluations.
Alibaba (Tongyi Lab)
Apache 2.0
Medium
en, zh, ja, ko, fr, de, it, es
Ha'e
Dia TTS
Estándar
Dia by Nari Labs is a 1.6B parameter text-to-speech model designed specifically for generating multi-speaker dialogue. It can produce natural-sounding conversations between two speakers with appropriate turn-taking, prosody, and emotional expression. Dia is perfect for creating podcast-style content, audiobook dialogues, and interactive conversational AI.
Nari Labs
Apache 2.0
Medium
en
No
Parler TTS
Estándar
Parler TTS is a text-to-speech model that uses natural language voice descriptions to control the generated speech. Instead of selecting from preset voices, you describe the voice you want (e.g., "a warm female voice with a slight British accent, speaking slowly and clearly") and Parler generates speech matching that description. This makes it uniquely flexible for creative applications.
Hugging Face
Apache 2.0
Medium
en
No
IndexTTS-2
Estándar
IndexTTS-2 is an advanced text-to-speech system that excels at zero-shot voice synthesis with fine-grained emotion control. It can generate speech with specific emotional tones like happy, sad, angry, or fearful without requiring emotion-specific training data. The model uses emotion vectors to precisely control the emotional expression of generated speech.
Index Team
Bilibili Model License
Medium
en, zh
Ha'e
Spark TTS
Estándar
Spark TTS by SparkAudio is a text-to-speech model that combines voice cloning with controllable emotion and speaking style. Using just 5 seconds of reference audio, it can clone a voice and then generate speech with different emotions, speeds, and styles while maintaining the cloned voice identity. Spark TTS uses a prompt-based control system.
SparkAudio
CC BY-NC-SA 4.0
Medium
en, zh
Ha'e
GPT-SoVITS
Estándar
GPT-SoVITS combines GPT-style language modeling with SoVITS (Singing Voice Inference via Translation and Synthesis) for powerful few-shot voice cloning. With as little as 5 seconds of reference audio, it can accurately clone a voice and generate new speech while preserving the speaker's unique characteristics. It excels at both speaking and singing voice synthesis.
RVC-Boss
MIT
Slow
en, zh, ja, ko
Ha'e
Orpheus
Estándar
Orpheus is a large-scale text-to-speech model that achieves human-level emotional expression. Trained on over 100,000 hours of diverse speech data, it excels at generating speech with natural emotions, emphasis, and speaking styles. Orpheus can produce speech that is virtually indistinguishable from human recordings.
Canopy Labs
Llama 3.2 Community
Medium
en
No
Qwen3 TTS
Estándar
Qwen3-TTS is a 1.7 billion parameter text-to-speech model from Alibaba's Qwen team. It supports two modes: preset voices with emotion control (9 speakers), and a unique voice design mode where you describe the voice you want in natural language. It covers 10 languages with high expressiveness and natural prosody.
Alibaba (Qwen)
Apache 2.0
Medium
en, zh, ja, ko, de, fr, ru, pt, es, it
No
VieNeu-TTS-v2
Estándar
VieNeu-TTS-v2 is a 300M parameter Vietnamese-first TTS model trained on 10,000+ hours of bilingual data. It supports seamless en-vi code-switching, 7 preset voices spanning Northern and Southern accents, and instant voice cloning from 3-5 seconds of reference audio. Runs entirely on CPU via GGUF Q4 inference + ONNX audio decoder — no GPU needed, generations finish in ~7 seconds. Built on a Qwen3 backbone.
Phạm Nguyễn Ngọc Bảo
Apache 2.0
Fast
vi, en
Ha'e
Chatterbox Turbo
Estándar
Chatterbox Turbo by Resemble AI is a 350M parameter upgrade to Chatterbox, delivering up to 6x real-time speed with sub-200ms latency. It supports paralinguistic tags like [laugh], [cough], and [chuckle] directly in text. Includes Perth watermarking on all generated audio for provenance tracking.
Resemble AI
MIT
Fast
en
Ha'e
VoxCPM
Estándar
VoxCPM 1.5 by OpenBMB is a novel tokenizer-free TTS model that operates in continuous space rather than discrete tokens. It produces high-fidelity 44.1kHz audio, supports zero-shot voice cloning from 3-10 seconds, and maintains consistency across paragraphs. Cross-language cloning lets you apply an English voice to Chinese speech and vice versa.
OpenBMB
Apache 2.0
Fast
en, zh
Ha'e
VibeVoice
Estándar
VibeVoice from Microsoft generates long-form speech up to 90 minutes with support for 4 simultaneous speakers, making it ideal for podcasts and dialogues. The Realtime 0.5B variant achieves ~300ms latency for interactive use. Supports speaker tags for multi-turn dialogue generation.
Microsoft
MIT
Fast
en, zh
No
CosyVoice3
Estándar
CosyVoice3 is the latest evolution from Alibaba's FunAudioLLM team. It features bi-streaming inference with ~150ms latency, instruction-based control for emotion/speed/volume, and improved speaker similarity for zero-shot cloning. Supports 9 languages plus 18 Chinese dialects. RL-tuned variant delivers state-of-the-art prosody.
Alibaba (FunAudioLLM)
Apache 2.0
Fast
en, zh, ja, ko, de, es, fr, it, ru
Ha'e
NAMAA Saudi TTS
Estándar
NAMAA Saudi TTS is a Saudi Arabic fine-tune of Resemble AI's ChatterboxMultilingual. Trained by NAMAA Space on authentic Saudi-dialect speech, it produces natural Modern Standard Arabic and Saudi colloquial pronunciation that generic multilingual models cannot match. Inherits Chatterbox's zero-shot voice cloning and emotion control via reference audio prompts. The first open-weights Arabic TTS deployed on TTS.ai.
NAMAA Space
MIT
Medium
ar
Ha'e
Darwin TTS
Estándar
Darwin-TTS-1.7B-Cross by FINAL-Bench is a research variant of Qwen3-TTS-1.7B where 84 talker-FFN tensors (8.6%) are blended at α=3% with the matching tensors from Qwen3-1.7B-Base. The blend is built without retraining and produces noticeably crisper cross-lingual voice cloning across Korean, English, Japanese, and Chinese. Operates in zero-shot voice-clone mode (3 seconds reference audio).
FINAL-Bench
Apache 2.0
Medium
en, ko, ja, zh
Ha'e
MOSS-TTSD
Estándar
MOSS-TTSD v1.0 from OpenMOSS is a 7B dialogue text-to-speech model that continues conversations from a short audio prompt. Supports up to 5 simultaneous speakers via [S1]/[S2] tags, zero-shot voice cloning from 3-10s reference audio, and up to 60 minutes of coherent multi-turn dialogue across 20 languages. Distinct from MOSS-TTS — TTSD is specialized for podcast/audiobook/dubbing workflows.
OpenMOSS
Apache 2.0
Medium
en, zh
Ha'e
Chatterbox
Premium
Chatterbox by Resemble AI is a cutting-edge zero-shot voice cloning model. It can replicate any voice from a single audio sample with remarkable accuracy, capturing not just the timbre but also the speaking style and emotional nuances. Chatterbox also features fine-grained emotion control, allowing you to adjust the emotional tone of the generated speech independently from the voice identity.
Resemble AI
MIT
Medium
en
Ha'e
4GB
4x
Tortoise TTS
Premium
Tortoise TTS is an autoregressive multi-voice text-to-speech system that prioritizes audio quality over speed. It uses DALL-E-inspired architecture to generate highly natural speech with excellent prosody and speaker similarity. While slower than many alternatives, Tortoise produces some of the most realistic synthetic speech available in the open-source ecosystem.
James Betker
Apache 2.0
Slow
en
Ha'e
8GB
4x
StyleTTS 2
Premium
StyleTTS 2 achieves human-level TTS synthesis by combining style diffusion with adversarial training using large speech language models. It generates the most natural sounding speech among single-speaker models, rivaling human recordings. StyleTTS 2 uses diffusion-based style modeling to capture the full range of human speech variation.
Columbia University
MIT
Medium
en
No
4GB
4x
OpenVoice
Premium
OpenVoice by MyShell.ai enables instant voice cloning with granular control over voice style, emotion, accent, rhythm, pauses, and intonation. It can clone a voice from a short audio clip and generate speech in multiple languages while maintaining the speaker identity. OpenVoice also functions as a voice converter, allowing real-time voice transformation.
MyShell.ai / MIT
MIT
Medium
en, zh, ja, ko, fr, es
Ha'e
4GB
4x
Sesame CSM
Premium
Sesame CSM (Conversational Speech Model) is a 1 billion parameter model designed specifically for generating conversational speech. It models the natural patterns of human conversation including turn-taking timing, backchannel responses, emotional reactions, and conversational flow. CSM generates audio that sounds like a natural human conversation rather than synthetic speech.
Sesame
Apache 2.0
Slow
en
No
8GB
4x
Modelo comparación rehegua tabla
| Modelo | Desarrollador: | Ta'ãnga | Calidad: | Velocidad | Ñe'ẽ | Clonación ñe'ẽnguéra rehe | VRAM | Licencia: | Presupuesto | |
|---|---|---|---|---|---|---|---|---|---|---|
| Kokoro | Hexgrad | Free | Fast | 8 | 1.5GB | Apache 2.0 | Libre | Ojeporu | ||
| Piper | Rhasspy | Free | Fast | 35 | 0 (CPU only) | MIT | Libre | Ojeporu | ||
| VITS | Jaehyeon Kim et al. | Free | Fast | 11 | 1GB | MIT | Libre | Ojeporu | ||
| MeloTTS | MyShell.ai | Free | Fast | 6 | 0.5GB (GPU optional) | MIT | Libre | Ojeporu | ||
| Bark | Suno | Standard | Slow | 13 | 5GB | MIT | 2 | Ojeporu | ||
| Bark Small | Suno | Standard | Medium | 13 | 2GB | MIT | 2 | Ojeporu | ||
| CosyVoice 2 | Alibaba (Tongyi Lab) | Standard | Medium | 8 | 4GB | Apache 2.0 | 2 | Ojeporu | ||
| Dia TTS | Nari Labs | Standard | Medium | 1 | 4GB | Apache 2.0 | 2 | Ojeporu | ||
| Parler TTS | Hugging Face | Standard | Medium | 1 | 4GB | Apache 2.0 | 2 | Ojeporu | ||
| IndexTTS-2 | Index Team | Standard | Medium | 2 | 4GB | Bilibili Model License | 2 | Ojeporu | ||
| Spark TTS | SparkAudio | Standard | Medium | 2 | 4GB | CC BY-NC-SA 4.0 | 2 | Ojeporu | ||
| GPT-SoVITS | RVC-Boss | Standard | Slow | 4 | 6GB | MIT | 2 | Ojeporu | ||
| Orpheus | Canopy Labs | Standard | Medium | 1 | 4GB | Llama 3.2 Community | 2 | Ojeporu | ||
| Chatterbox | Resemble AI | Premium | Medium | 1 | 4GB | MIT | 4 | Ojeporu | ||
| Tortoise TTS | James Betker | Premium | Slow | 1 | 8GB | Apache 2.0 | 4 | Ojeporu | ||
| StyleTTS 2 | Columbia University | Premium | Medium | 1 | 4GB | MIT | 4 | Ojeporu | ||
| OpenVoice | MyShell.ai / MIT | Premium | Medium | 6 | 4GB | MIT | 4 | Ojeporu | ||
| Qwen3 TTS | Alibaba (Qwen) | Standard | Medium | 10 | 7GB | Apache 2.0 | 2 | Ojeporu | ||
| VieNeu-TTS-v2 | Phạm Nguyễn Ngọc Bảo | Standard | Fast | 2 | CPU | Apache 2.0 | 2 | Ojeporu | ||
| Sesame CSM | Sesame | Premium | Slow | 1 | 8GB | Apache 2.0 | 4 | Ojeporu | ||
| Chatterbox Turbo | Resemble AI | Standard | Fast | 1 | 2GB | MIT | 2 | Ojeporu | ||
| VoxCPM | OpenBMB | Standard | Fast | 2 | 4GB | Apache 2.0 | 2 | Ojeporu | ||
| Kani TTS 2 | NineNineSix | Free | Fast | 1 | 3GB | Apache 2.0 | Libre | Ojeporu | ||
| OuteTTS | OuteAI | Free | Fast | 1 | 2GB | Apache 2.0 | Libre | Ojeporu | ||
| VibeVoice | Microsoft | Standard | Fast | 2 | 4GB | MIT | 2 | Ojeporu | ||
| Pocket TTS | Kyutai | Free | Fast | 2 | 1GB | MIT | Libre | Ojeporu | ||
| Kitten TTS | KittenML | Free | Fast | 1 | 0GB | Apache 2.0 | Libre | Ojeporu | ||
| CosyVoice3 | Alibaba (FunAudioLLM) | Standard | Fast | 9 | 4GB | Apache 2.0 | 2 | Ojeporu | ||
| NAMAA Saudi TTS | NAMAA Space | Standard | Medium | 1 | 6GB | MIT | 2 | Ojeporu | ||
| Darwin TTS | FINAL-Bench | Standard | Medium | 4 | 7GB | Apache 2.0 | 2 | Ojeporu | ||
| MOSS-TTSD | OpenMOSS | Standard | Medium | 2 | 12GB | Apache 2.0 | 2 | Ojeporu | ||
| Ming-Omni TTS | inclusionAI | Free | Medium | 2 | 3GB | Apache 2.0 | Libre | Ojeporu | ||
| MOSS-TTS Nano | OpenMOSS | Free | Fast | 11 | 2GB | Apache 2.0 | Libre | Ojeporu |
Plataforma IA ñeikumby ñe'ẽ'arã tuichavéva
Mba'érepa oiporavo TTS.ai ñe'ẽjoaju ñeikumbyrã?
TTS.ai ombojoaju umi modelo ñe'ẽ-gui-ñe'ẽ-pe g̃uarã fuente abierto-pegua iporãvéva arapy tuichakue peteĩ plataforma añónte, ndahasýi ojeporu hag̃ua. Umi servicio propiedad-peguápe ojoavýva ombohape hag̃ua peteĩ motor ñe'ẽ rehegua añónte, TTS.ai ome'ẽ acceso hetave 20 modelo-pe g̃uarã umi laboratorio de investigación ha'evéva apytépe, oikehápe Coqui, MyShell, Amphion, NVIDIA, Suno, HuggingFace, Tsinghua Universidad ha hetave.
Opaite modelo hína código abierto MIT, Apache 2.0 térã ambue licencia permisiva rupive, ombohapehápe ikatu hag̃uaicha oguereko derecho comercial oipuru hag̃ua umi sonido ojehupytýva nde proyecto-pe. Oikotevẽramo peteĩ síntesis ipya'e ha pya'egua umi aplicación tiempo real-pe g̃uarã térã peteĩ salida estudio-pegua calidad-pe g̃uarã audiolibro ha podcast-pe g̃uarã, TTS.ai oguereko modelo oĩporãva oimeraẽ jeporurã.
Modelos libres, ndoikotevẽi peteĩ cuenta
Oñepyrũ pya'e mbohapy modelo TTS-pe: Piper (hypy'ũva, ipya'evéva), VITS (síntesis neuronal calidad-py yvatevéva) ha MeloTTS (apopyrã heta ñe'ẽme). Ndojehechavéima registro, tarjeta de crédito, generación-kuéra límite. Umi modelo libre oykeko inglés ha ambue ñe'ẽnguéra, oguerekóva salida natural ha oguerekóva sonido oikéva heta aplicación-pe g̃uarã.
Procesamiento acelerado GPU rupive
Opaite modelo TTS oiko GPU NVIDIA-pe oñemohendáva, generación pya'eve ha katui hag̃ua. Umi modelo gratuito katuínte omoheñói ñehendurã 2 segundo sa'ive aja. Umi modelo estándar, taha'e Kokoro, CosyVoice 2 ha Bark, oguereko peteĩ promedio 3 ha 5 segundo rupi. Umi modelo premium calidad ijyvatevéva, taha'e Tortoise ha Chatterbox, omoheñói ñehendurã 5 ha 15 segundo rupi, ojehechahápe ñe'ẽnguéra ipukukue.
30+ ñe'ẽ ojeykeko
Oñemoheñói ñe'ẽnguéra hetave 30 ñe'ẽme, oikehápe inglés, español, francés, alemán, italiano, portugués, chino, japonés, coreano, árabe, hindi, ruso ha hetave. Heta modelo oykeko ñe'ẽnguéra ñembojoaju, he'iséva ikatuha omoheñói ñe'ẽnguéra peteĩteĩ peteĩ ñe'ẽme ndojeikuaa'ỹva. CosyVoice 2 ha GPT-SoVITS ojehecharamo clonación ñe'ẽnguéra ñembojoajurã.
API oñembosako'iporãva umi desarrollador-pe g̃uarã
Oike TTS.ai-pe apopyrãkuérape API REST OpenAI-pe ojokupytýva rupive. Peteĩ punto final opaite 20+ modelo-pe g̃uarã. Python, JavaScript, cURL ha Go SDK. Apopyrã tiempo real-pe g̃uarã. Procesamiento por lotes contenido generación tuichaháicha-pe g̃uarã. Webhooks notificación asincrónica-pe g̃uarã. API jeike oike hína oimeraẽ plan-pe, oikehápe avei libre.
Pregunta ojejapóva py'ỹinte
Mba'épa ikatu ñambohape? Tuichave ñemoneĩ oipytyvõta ñamoambue hag̃ua umi apañuãi.
Oñepyrũ conversi'ỹva ñe'ẽ'ỹme
Ojoaju hetaiterei creador-pe oiporúvo TTS.ai. Oñemoinge 15000 caractere'ỹre peteĩ cuenta pyahu rupive. Oĩ modelo'ỹre ojeiporukuaáva ndojehechakuaái rupi.