Okubhaliweyo ukuya kuSpeechName Abathunywa be-AI I-Music Umbhalo we-audio I-Voice Cloning Uyilo lwesandi I-Marketplace I-More

Umbhalo ovulekileyo ukuya kwimodeli yokuthetha

Imodeli nganye ye-TTS kwinkqubo yethu ivela kwi-open source enelayisensi elungele ishishini. MIT, Apache 2. 0 — akukho lungelo lokutshixa, akukho mda wokusetyenziswa, akukho xabiso lelayisensi elingaziwayo. Sebenzisa ngokudibanisa ne-API yethu, okanye ubeke ngokwakho kwinkqubo yakho yolawulo olupheleleyo.

Ikhowudi evulekileyo Ilayisensi ye-MIT Apache 2. 0 I-self-hosting GitHub
Umhleli we-TTS opheleleyo Uxwebhu lwe-API

Zama Ngoku

0/500
Ikhululekile nge Kokoro, Piper, VITS, MeloTTS
Isandi sakho esivelisweyo siza kuvela apha
Iveliswe
Layishela phantsi egronjiweyo
Uthando TTS.ai? Nceda utshele abalandeli bakho!
Vula umhleli we-TTS opheleleyo

IiNkqubo zeTTS ezivulekileyo

Iimodeli ezivulekileyo ze-source zibalulekile njani kwiprojekthi zakho

Zonke i-Open Source Licensed

Imodeli nganye kwi TTS.ai isebenzisa ilayisensi evulekileyo evulekileyo. Akukho bhokisi emnyama esemthethweni, akukho mthengisi otshixiwe, akukho xabiso lelayisensi elingalindelekanga.

MIT / Apache 2. 0

Iimodeli zilayisensiwe phantsi kwe MIT okanye i-Apache 2.0, ilayisensi evulekileyo evulekileyo. Sebenzisa ngokurhweba, guqula, phinda unikezele — akukho mda.

I-self-hosting

Layisha ezantsi nayiphi na imodeli kwaye uyiqhube kwihardware yakho. Ulawulo olupheleleyo kwidata yakho, ukulinda, kunye nenkqubo yokusebenza. Akukho xhomekeko kwicloud efunekayo.

GPU elungelelanisiweyo

Iimodeli zilungelelaniswe kakuhle kwi-NVIDIA GPUs ene-CUDA inkxaso. I-Piper isebenza kwi-CPU kuphela. Iimodeli ezininzi zifuna i-2-8GB VRAM yokwahlula ngokufanelekileyo.

Iinkqubo ezixhaswa

Iindawo ezisebenzayo ezivulekileyo zigcina kwaye ziphucula ezi modeli. Iingxelo ziyavuya — thumela iibugs, ukuphuculwa, kunye neelizwi ezintsha kwi GitHub.

I-Commercial Use OK

Zonke iimodyuli zivumela ukusetyenziswa korhwebo phantsi kwelayisensi zabo. Yenza iimveliso, thengisa iinkonzo, kwaye yenza imixholo yorhwebo ngaphandle kwee-royalties okanye iindleko zokusetyenziswa.

I-Open Source Model Catalog yethu

Imodeli nganye, ilayisensi yayo, nento eyenza kakuhle

KokoroKokoro

Free

Lightweight 82M parameter model delivering studio-quality speech with blazing-fast inference.

Fast 5/5

Elungileyo ku: Apache 2. 0 - umgangatho olungileyo wemodeli ekhululekileyo, 82M params, kulula ukuphatha ngokwakho

Zama Kokoro

PiperPiper

Free

A fast, local neural text to speech system optimized for Raspberry Pi and embedded devices.

Fast 3/5

Elungileyo ku: MIT - CPU- kuphela, ilungile kwizixhobo zesiphelo kunye nokuhombisa okuzenzekelayo okufakwe ngaphakathi

Zama Piper

VITSVITS

Free

Conditional variational autoencoder with adversarial learning for end-to-end text-to-speech.

Fast 3/5

Elungileyo ku: MIT — uyilo lwesiseko olusetyenziswa ziimodyuli ezininzi eziphantsi kolwandle

Zama VITS

BarkBark

Standard

Transformer-based text-to-audio model that generates realistic speech, music, and sound effects.

Slow 4/5

Elungileyo ku: I-MIT — iimpawu ezikhethekileyo zokuveliswa kwesandi ngaphezulu kwe-TTS eqhelekileyo

Zama Bark

Tortoise TTSTortoise TTS

Premium

Multi-voice text-to-speech focused on quality with autoregressive architecture.

Slow 5/5 I-Voice Cloning

Elungileyo ku: Apache 2. 0 - umgangatho ophezulu, unikezelo olufundelwe ngokubanzi

Zama Tortoise TTS

OpenVoiceOpenVoice

Premium

Instant voice cloning with granular control over style, emotion, and accent.

Medium 4/5 I-Voice Cloning

Elungileyo ku: MIT - uklonelo lwesandi somthombo ovulekileyo nolawulo lwesitayile esithe nkqo

Zama OpenVoice

Indlela Yokusebenzisa i Open Source TTS

Sebenzisa i-API yethu ekhoyo okanye uqhube iimodeli ngokwakho

1

Khangela iimodyuli ezivulekileyo

Khangela i-catalog yethu ye-20+ ye-open-source TTS models. Iphepha ngalinye lemodeli libonisa ilayisensi, uyilo, ubukhulu, kunye neemfuno zokuhombisa ngokwazo.

2

Zama kwiBhrabhseri Yakho

Uvavanyo lwemodeli ngqo kwi TTS.ai ngaphandle kokufaka nantoni na. Amaseva ethu e GPU aphatha uqhubekeko ukuze ukwazi ukuvavanya umgangatho phambi kokuba ubeke isandla ekusebenzeni ngokuzenzekelayo.

3

I-Auto-Host okanye Sebenzisa i-API yethu

Uhlobo lwe-clone lwe-repos yemodeli ukusuka kwi-GitHub kwaye uqhube ngaphakathi, okanye sebenzisa i-API yethu ekhoyo yokwenziwa. Ukuhlala ngokwakho kunika ulawulo olupheleleyo; i-API yethu ibonelela ngenkqubo elawulwayo.

4

Yenza Isicelo Sam

Yongeza i-TTS kwimveliso yakho usebenzisa iimodyuli ezibekwe ngokwazo okanye i-REST API yethu. Zonke iimodyuli zisetyenziswa ngokurhweba ngaphandle kweemali zokufaka ilayisensi okanye iirhafu.

Uthelekiso lwelayisensi

Zonke iimodyuli kwi-TTS.ai zisebenzisa iileyisensi ezivulekileyo ezilungele urhwebo

Imodeli Ilayisensi Ukusetyenziswa kwentengiso Utshintsho I-Host-Ezimeleyo Unikezelo
Kokoro Apache 2.0 Ifuneka
Piper MIT Ekunokukhethwa kuko
VITS MIT Ekunokukhethwa kuko
MeloTTS MIT Ekunokukhethwa kuko
Chatterbox MIT Ekunokukhethwa kuko
Tortoise TTS Apache 2.0 Ifuneka
StyleTTS 2 MIT Ekunokukhethwa kuko
OpenVoice MIT Ekunokukhethwa kuko
Sesame CSM Apache 2.0 Ifuneka
Orpheus Llama 3.2 "Built with Llama"
Try These Models Free

Ukugcina ngokuzenzekelayo vs Ukugcina i-API

Uqhube iimodeli ngokwakho okanye usivumele siphatha iinkxaso-

I-Host Ebonakalayo Kwizixhobo Zokwandisa

Imodeli nganye kwi-TTS.ai ifumaneka njengeprojekthi yomthombo ovulekileyo kwi-GitHub okanye i-Hugging Face. Layisha ezantsi iintonga, ufake izimeleyo, kwaye uqhube ukuqonda kwi-GPU yakho. Unayo ulawulo olupheleleyo kwi-latency, ubumfihlo, kunye nokulinganisela.

  • Ukhuseleko lwedata olupheleleyo — isandi asiyi kushiya iseva yakho
  • Akukho xabiso lesicelo ngasinye emva kokufaka ngokuzenzekelayo
  • Ulawulo oluzenzekelayo lwe-data yakho
  • Ifuna i-hardware ye-GPU (i-NVIDIA icetyiswa)
  • Uphatha uhlaziyo, ukulinganisa, kunye nokuxhomekeka

Sebenzisa i TTS.ai Hosted API

Fumana ukufikelela ngokukhawuleza kuzo zonke iimodeli ezingama-20+ nge-REST API enye. Siphatha unikezelo lwe-GPU, uhlaziyo lwemodeli, ulawulo lofolo, kunye nokunyuka. Isitshixo se-API esinye sikunika ukufikelela kwimodeli nganye - akukho mfuneko yokulawula unikezelo oluhlukileyo.

  • Akukho zixhobo zekhompyutha ze-GPU ezifunekayo
  • Zonke iimodeli ezingama-20+ zisebenzisa i-API enye
  • Imodeli ekhawulezayo yokugqiba kunye nokuphunyezwa
  • 99.9% yexesha elisebenzayo
  • Uhlawula kuphela oku kusetyenziswa
Use Hosted API Instead

Iqala ngokukhawuleza: API okanye Umququzeleli Ozimeleyo

Sebenzisa i-API yethu ekhoyo, okanye ufake i-Kokoro kwindawo yakho kwimizuzu

Ukhetho 1: TTS.ai Ikhowudi ebhaliweyo ye API Elula 
import requests

response = requests.post("https://api.tts.ai/v1/tts", json={
    "text": "Open source TTS with a simple API.",
    "model": "kokoro",
    "voice": "af_heart",
    "format": "wav"
}, headers={"Authorization": "Bearer YOUR_API_KEY"})

with open("output.wav", "wb") as f:
    f.write(response.content)
Ukhetho 2: I-Host- Yakho- Nge-pip Ulawulo olupheleleyo 
# Install Kokoro locally
pip install kokoro

# Generate speech on your own GPU
import kokoro

pipeline = kokoro.KPipeline(lang_code="a")
generator = pipeline("Hello from your own server!", voice="af_heart")
for i, (gs, ps, audio) in enumerate(generator):
    kokoro.save(audio, f"output_{i}.wav")
View API Documentation

I-Open Source, ixabiso elifanelekileyo

I-API yethu ekhoyo isenza ukuba i-open-source TTS ifikeleleke ngaphandle kokuphatha ii-GPUs.

Umphakamo okhululekileyo

$0

15,000 iimpawu kwi-signup

  • 4 iimodyuli ezivulekileyo ezikhululekileyo
  • Akukho ubhaliso lokusetyenziswa okusisiseko
  • Ukusetyenziswa korhwebo kuvunyelwe

Isiqalisi

$9

500,000 iimpawu/inyanga

  • Zonke iimodeli ezivulekileyo ezingaphezu kwe-20
  • Ukuphinda usebenzise ilizwi
  • Ufikelelo lwe-API

I-Pro

$29

2,000,000 characters/month

  • Uqhubekeko lwe-GPU oluphambili
  • Zonke iimodeli eziphezulu
  • Uxhaso lweshishini
Ixabiso elipheleleyo

Imibuzo ebuzwa rhoqo

Imibuzo ebuzwa rhoqo malunga nombhalo ovulekileyo womthombo wokuthetha

Ewe. Imodeli nganye kwi-TTS.ai isebenzisa ilayisenisi evulekileyo evulekileyo — nokuba yi-MIT okanye i-Apache 2.0. Siyikhupha ngokukodwa iimodyuli ezinelayisensi ezithintelayo (njenge-Coqui's CPML okanye i-CC-BY-NC engarhwebiyo). Ungaqinisekiswa ilayisenisi yemodeli nganye kwi-GitHub yayo.

Zonke ziilayisensi ezivulekileyo ezivumela ukusetyenziswa korhwebo, utshintsho, kunye nokuphinda kunikezelwe. I-Apache 2. 0 idibanisa izivumelwano ezicacileyo zepatent kwaye ifuna ukuchaza utshintsho ukuba uguqula ikhowudi. I-MIT ilula ngeemfuno ezincinci. Zonke zisebenza kakuhle.

Ewe. Imodeli nganye inokugcinwa ngokuzimela. Khuphela imodeli yendawo yokugcina i-GitHub, ufake izimeleyo, khuphela ezantsi imodeli yomthwalo, kwaye uqhube ukuqonda. Sinika uxwebhu lweemodeli nganye ezifuna ukugcinwa ngokuzimela kubandakanya i-GPU, i-RAM, kunye noguqulelo lwe-Python.

Iimfuno ziyahluka ngokwemodeli. I-Piper ayifuni i-GPU (i-CPU kuphela). I-Kokoro ne-MeloTTS zifuna i-1-2GB ye-VRAM. Iimodeli ezininzi eziqhelekileyo zifuna i-4GB ye-VRAM. I-Tortoise ne-Sesame CSM zifuna i-8GB. I-NVIDIA RTX 3060 (12GB) ingaqhuba iimodyuli ezininzi ngokukhululekileyo.

Ewe. Iilayisensi zomthombo ovulekileyo zivumela utshintsho kuquka nokuhlengahlengisa. Iimodeli ezifana ne-GPT-SoVITS ne-Bark zibonelela ngeeskripthi zokuhlengahlengisa. Ungaziqeqesha iimodeli kwidata yakho yelizwi ukuze wenze ilizwi elikhethekileyo okanye ukuphucula ukusebenza kweelwimi ezithile.

Iimodeli eziphezulu ezivulekileyo (iKokoro, iStyleTTS 2, iChatterbox) ngoku zifana okanye zingaphezulu kweenkonzo zentengiso ezinjengeElevenLabs neGoogle TTS kwiimpawu zomgangatho. Inzuzo enkulu yenkonzo yentengiso kukulawulwa kwenkqubo yokwakha kunye noxhaso, hayi umgangatho wesandi.

Sisele sizikhuphe. XTTS/XTTS-v2 (i-Coqui's CPML - engarhwebiyo), F5-TTS (i-CC-BY-NC - engarhwebiyo), ne-Higgs-v2 (i-Boson License - ethintelayo) zonke zasuswa. Yonke imodeli kwi-TTS.ai iqinisekisiwe ukuba ikhuselekile kwi-intengiso.

Ewe. Iimodeli ezininzi zixhasa uncedo lweqela nge-GitHub. Ungathumela iingxelo zegciwane, ukurekhodwa kwesandi kwezilwimi ezintsha, ukuphuculwa kwekhowudi, kunye noxwebhu. Khangela imodeli nganye ye-GitHub yokugcina imiyalelo yoncedo kunye nengxaki esebenzayo.

Faka iimodeli kwisicelo kwaye ulayishe xa ungekho esebenzayo ukudibanisa inkumbulo ye-GPU. Iseva yethu ye-GPU iqhuba iimodeli ezingama-20 + kwi-4x Tesla P40 (i-96GB ye-VRAM) usebenzisa ukufaka okukhawulezayo. Ukugcina ngokwakho, i-24GB GPU enye inokuncedisa iimodeli ezingama-3-5 ngokufanayo.

Iimodeli ezininzi zinika imifanekiso yeDocker okanye iifayile zeDocker. Ukuqhuba iimodeli ezininzi, ungakha isicwangciso seDocker esizithandayo nge-NVIDIA Container Toolkit ye-GPU. Uyilo lweseva ye-API yethu lunokusetyenziselwa ukubhekisa kwinkqubo.

Iimodeli ezininzi zifuna i-Python 3.10-3.12. I-Coqui TTS (VITS) ifuna i-Python 3.11. Sicebisa i-Python 3.12 kwiimodeli ezininzi. Khangela i-requirements.txt yemodeli nganye ukuqinisekisa ukulungelelaniswa kohlobo.

Ewe. I-MIT ne-Apache 2.0 ilayisensi zivumela ngokucacileyo ukusetyenziswa korhwebo. Ungayila iimveliso ze-SaaS, iinkqubo zeselfowuni, imidlalo, kunye neenkonzo usebenzisa ezi modeli ngaphandle kweemali zokufaka ilayisensi, iirhafu, okanye iimfuno zokwabelana (ukuba ukuwabelana kuthandwa).
5.0/5 (1)

Yintoni esinokuyilungisa? Ulwazi lwakho olufunyenweyo lunceda silungise iingxaki.

Zama i Open Source TTS Namhlanje

20+ iimodyuli ezivulekileyo, zonke zilayisensiwe ngentengiso. Sebenzisa i-API yethu okanye i-self-host - ukhetho luya kuwe.

Ubhaliso simahla Bonisa ixabiso