Text to SpeechQuery Oyang'anira a AI Mtima Kulankhula kwa Malemba Chizindikiro cha mawu Kujambula kwa mawu Malo Ogulitsa Kuwonjezera
Kutumiza Zidziwitso za Zifukwa

Open Source Text to Speech Models

Mtundu uliwonse wa TTS papulatifomu yathu ndi waulere, ndipo ambiri amagwiritsa ntchito MIT kapena Apache 2.0. Mtundu ndi mawu ena ali ndi chilolezo chogwiritsa ntchito zosapanga ndalama zokha ndipo amadziwika kuti "Personal and non-commercial use only". Mutha kuwagwiritsa ntchito pogwiritsa ntchito API yathu yokhala ndi ma API, kapena kukhazikitsa iwo pazinthu zanu zokha ndi kuwongolera kwathunthu.

Zolemba Zotsegulidwa MIT License Apache 2.0 Kukhala ndi Mlengi GitHub
Full TTS wolemba Mabuku a API

Yambitsani Tsopano

0/500
Free ndi Kokoro, Piper, VITS, MeloTTS
Zina zanu zopangidwa ndi mawu zidzawonekera pano
Zopangidwa
0:00
Kutsitsa
Kukonda TTS.ai? udzauza anzanu!
Atsegula TTS wolemba

Open Source TTS Zothandiza

Nchifukwa chiyani mapangidwe a otsegulira amakhudza mabizinesi anu

Zonse Open-Source Licensed

Mtundu uliwonse pa TTS.ai ndi otsegulira, ndi m'munsimu ndi liwiro la TTS.ai.com.

Otsegula Source Licenses

Mamodeli ambiri amagwiritsa ntchito MIT kapena Apache 2.0. Palibe amene amagwiritsa ntchito zina, ndipo mamodeli kapena mawu omwe amagwiritsa ntchito ntchito zosagulitsa amadziwika kuti "Personal and non-commercial use only".

Kukhala ndi Mlengi

Mukhoza kutsitsa chilichonse cha mtundu uliwonse ndikuchigwiritsa ntchito pa zida zanu. Mukhoza kuwongolera kwathunthu pa data yanu, kulephera, komanso chitetezo.

GPU yosinthidwa

Mapangidwe amasinthidwa kuti azigwirizana ndi NVIDIA GPUs ndi CUDA. Piper imagwira ntchito pa CPU yokha. Mapangidwe ambiri amafunikira 2-8GB VRAM kuti azitha kuzindikira bwino.

Kugwirizana

Mabungwe otsegulira otsegulira amasamalira ndi kukulitsa machitidwewa. Kuphatikiza kwaulere - kutumiza ma bugs, kukulitsa, ndi mawu atsopano pa GitHub.

Commercial License Opezeka

Mtengo wolipira wa $5/mwezi umabweretsa chilolezo chogulitsa kwa zomwe mumapanga, popanda ndalama zolipira kapena ndalama zogwiritsa ntchito; chigawo choulere ndi chogwiritsa ntchito pamodzi. Models ndi mawu omwe amadziwika kuti "Personal and non-commercial use only" amagawidwa.

Timaperekanso Open Source Model Catalog

Gawo lililonse la mtundu, lisensi yake, ndi zomwe zimachita bwino

KokoroKokoro

Free

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

Mofulumira 5/5

Oyenera kwa: Apache 2.0 - yabwino kwambiri yaulere yaulere, 82M params, yosavuta kuyendera

Phunzirani Kokoro

PiperPiper

Free

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

Mofulumira 3/5

Oyenera kwa: MIT - CPU yokha, yabwino kwa zida za edge ndi embedded self-hosting

Phunzirani Piper

VITSVITS

Free

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

Mofulumira 3/5

Oyenera kwa: MIT - Foundational architecture amagwiritsa ntchito ambiri downstream mafano

Phunzirani VITS

BarkBark

Standard

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

Mofulumira 4/5

Oyenera kwa: MIT - osiyana audio chitukuko zipangizo m'mphepete mwa standard TTS

Phunzirani Bark

Tortoise TTSTortoise TTS

Premium

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

Mofulumira 5/5 Chizindikiro cha mawu

Oyenera kwa: Apache 2.0 - kukula kwabwino kwambiri, kuphunzira kochuluka pamalamulo

Phunzirani Tortoise TTS

OpenVoiceOpenVoice

Premium

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

M'mawu 4/5 Chizindikiro cha mawu

Oyenera kwa: MIT - kutulutsa mawu otsegulira ndi kuwongolera kwamtundu wa granular

Phunzirani OpenVoice

Momwe Mungagwiritsire Ntchito Open Source TTS

Mutha kugwiritsa ntchito ma API athu kapena kuyendetsa mapangidwe anu okha

1

Kafukufuku wa Open Source Models

Pezani mnda wathu wa 20+ open-source TTS models.Kawirikawiri, tsamba la model limasonyeza chilolezo, chilankhulo, zofunikira, ndi zofunikira za self-hosting.

2

Pezani mu msakatuli wanu

Kuyesa chilichonse chopangidwa mwachindunji pa TTS.ai popanda kukhazikitsa chilichonse. GPU servers athu amasamalira kugwiritsira ntchito kuti muthe kuyerekezera mtunduwu pambuyo pokhazikitsa kutengerapo.

3

Self-Host kapena kugwiritsa ntchito API yathu

Clone model repos kuchokera ku GitHub ndikugwira ntchito pamalo, kapena kugwiritsa ntchito API yathu yoyang'anira kupanga.Self-hosting imapatsa kuwongolera kwathunthu; API yathu imapatsa njira yoyang'anira.

4

Kukhazikitsa wanu Application

Kuphatikiza TTS muzinthu zanu pogwiritsa ntchito ma modeli oyang'anira okha kapena REST API yathu. Chotsani chilolezo cha mtundu uliwonse wa model: ma modeli ndi mawu ena ndi ogwiritsira ntchito opanda phindu.

Kuyerekezera kwa License

License ya aliyense mtundu pa TTS.ai

Model License Kugwiritsa ntchito kwamalonda Kusintha Woyang'anira yekha Kugwirizana
Kokoro Apache 2.0 Zofunika
Piper MIT injini; mawu licenses kusintha Mawu ena okha Chosafunikira
VITS MIT Chosafunikira
MeloTTS MIT Chosafunikira
Chatterbox MIT Chosafunikira
Tortoise TTS Apache 2.0 Zofunika
StyleTTS 2 MIT Chosafunikira
OpenVoice MIT Chosafunikira
Sesame CSM Apache 2.0 Zofunika
Orpheus Llama 3.2 "Built with Llama"
Spark TTS CC BY-NC-SA 4.0 Kugwiritsa ntchito kokha kwa munthu ndi kopanda bizinesi Zofunika
OuteTTS CC BY-NC-SA 4.0 Kugwiritsa ntchito kokha kwa munthu ndi kopanda bizinesi Zofunika
Phunzirani izi Models kwaulere

Kukhala ndi Hosting vs Hosted API

Kuyendetsa mapangidwe anu nokha kapena tikuthandizeni kusamalira chitetezo

Self-Host pa Hardware yanu

Mtundu uliwonse wa TTS.ai ulipo ngati projekiti yotsegulidwa pa GitHub kapena Hugging Face. Mutha kutsitsa kunenepa, kukhazikitsa zotengera, ndi kuyendetsa chidziwitso pa GPU yanu.Muli ndi kuwongolera kwathunthu pa latency, privacy, ndi scaling.

  • Full deta privacy - audio sadzasiya seva yanu
  • Palibe ndalama za per-zofuna pambuyo pa setup yoyamba
  • Custom fine-tuning pa wanu okha deta
  • Amafunikira zida za GPU (NVIDIA imalimbikitsa)
  • Mungathe kuyendetsa zosintha, scaling, ndi kutengerapo

Kugwiritsa ntchito TTS.ai Hosted API

Tikuwongolera kukhazikitsa kwa GPU, kusinthidwa kwa mapangidwe, kuwongolera kwa queue, ndi kukulitsa. Chimodzi mwazinthu za API zimakupatsani mwayi wopeza mapangidwe onse - palibe chifukwa chofunikira kuyendetsa kukhazikitsidwa kosiyana.

  • Palibe zida za GPU zofunikira
  • Zomwe 20 + zojambula ndi imodzi API
  • Kusintha kwa machitidwe ndi kuwonjezeka kwa machitidwe
  • 99.9% uptime ndi chitetezo chokwanira
  • Pezani ndalama zokha za zomwe mumagwiritsa ntchito
Kugwiritsa ntchito Hosted API

Kuyambira kwaulere: API kapena Self-Host

Mutha kugwiritsa ntchito API yathu yokhala ndi masamba, kapena kukhazikitsa Kokoro m'malo mwake m'maola angapo

Njira 1: TTS.ai Hosted API Yabwino kwambiri 
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)
Chosankha 2: Self-Host ndi pip Kuwongolera kwathunthu 
# 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

Otsegulidwa Source, Zotsika mtengo

Timapereka API yathu yokhala ndi TTS yotsegulira popanda kuyendetsa GPUs.

Free Tier

$0

15,000 characters pa signup

  • 4 open-source zojambula zaulere
  • Simungalembetse pogwiritsa ntchito mfundo
  • Kugwiritsa ntchito kokha, kopanda bizinesi

Woyamba

$9

500,000 characters/mwezi

  • Zonse 20+ open-source mapangidwe
  • Kusintha kwa mawu
  • Kupeza kwa API

Pro

$29

2,000,000 characters/mwezi

  • Kugwiritsa ntchito GPU
  • Zomwe zimadziwika bwino
  • Kuthandizira kwa Enterprise
Pangani Full Pricing

Funso Lofunsidwa Kawirikawiri

Mafunso ofala kwambiri pa nkhani ya kusinthitsa malemba kukhala mawu

Ndikoyenera, zonse zojambula pa TTS.ai ndi zochokera m'manja, ndipo ambiri amagwiritsa ntchito MIT kapena Apache 2.0. Palibe ambiri omwe amagwiritsa ntchito ma licenses ena, kuphatikizapo ena omwe ali ndi ma licenses osati ogwiritsira ntchito malonda okha, omwe amadziwika kuti "Personal and non-commercial use only" mu wosankha mawu.

Apache 2.0 imawonjezera zopereka za patenti zosadziwika bwino ndipo imafunikira kufotokoza zosintha ngati musintha kodi. MIT ndi yosavuta ndi zofunikira zochepa. Ambiri ndi othandiza pabizinesi.

Yai. Kawirikawiri, maphunzirowa amapezeka pa GitHub, koma titha kukhazikitsanso maphunzirowa pa GitHub. Ndikofunika kuti mupange maphunziro anu pa GitHub, kukhazikitsa zofunikira, kutsitsa maphunziro, ndi kuyendetsa chidziwitso. Timapereka zidziwitso za zofunikira za maphunziro onse, kuphatikizapo GPU, RAM, ndi mtundu wa Python.

Mafunso amasiyana malinga ndi mtundu. Piper safunikira GPU (CPU yokha). Kokoro ndi MeloTTS amafunikira 1-2GB VRAM. Mapangidwe ambiri amafunikira 4GB VRAM. Tortoise ndi Sesame CSM amafunikira 8GB. A NVIDIA RTX 3060 (12GB) angagwiritsidwe ntchito bwino kwambiri.

Yes. Open-source licenses allow modification including fine-tuning. Models like GPT-SoVITS and Bark provide fine-tuning scripts. You can train models on your own voice data to create custom voices or improve performance for specific languages.

Top open-source mafano (Kokoro, StyleTTS 2, Chatterbox) tsopano kugwirizana kapena kupitilira malonda ntchito monga ElevenLabs ndi Google TTS mu quality benchmarks.The lalikulu vantaggio di servizi commerciali è gestita infrastruttura e supporto, non qualità audio.

Ena mwa iwo atayatsidwa: XTTS/XTTS-v2 (Coqui CPML, osagulitsa), F5-TTS (CC-BY-NC, osagulitsa) ndi Higgs-v2 (Boson License, woletsa). Mamodeli ndi mawu ena omwe alipo, monga Spark TTS, OuteTTS ndi mawu ambiri a Piper, ali ndi ufulu wogwiritsa ntchito osagulitsa; amadziwika kuti "Personal and non-commercial use only" mu wosankha mawu, choncho kugwiritsa ntchito mawu ena kwa ntchito zogulitsa.

Ndikofunika. Mamodeli ambiri amavomereza maphunziro a anthu kudzera pa GitHub. Mukhoza kutumiza zidziwitso za zifukwa, zolemba za mawu za maphunziro atsopano, kuwonjezera kofunikira, ndi zolemba. Sankhani GitHub repository ya model iliyonse kuti mudziwe malangizo a maphunziro ndi mavuto ogwira ntchito.

GPU yathu ikugwira ntchito ndi 20 + mapangidwe pa 4x Tesla P40 (96GB VRAM yonse) pogwiritsa ntchito kutsitsa kwa dynamic. Pogwiritsa ntchito 24GB GPU, mutha kuyendetsa ma 3-5 mapangidwe nthawi imodzi.

Mapangidwe ambiri amapatsa zithunzi za Docker kapena Dockerfiles. Kuti mugwiritse ntchito mapangidwe ambiri, mutha kupanga setup ya Docker yosinthidwa ndi NVIDIA Container Toolkit kuti mupite ku GPU.

Mapangidwe ambiri amafuna Python 3.10-3.12. Coqui TTS (VITS) amafunikira Python 3.11. Tikulimbikitsa Python 3.12 kwa mapangidwe ambiri. Onani requirements.txt ya mapangidwe onse kuti mudziwe mtundu wogwirizana.

Ngati mumangokhala, makamaka yeah: MIT ndi Apache 2.0 malayisensi amalola kugwiritsa ntchito malonda, ndipo ambiri a mamodeli apa amagwiritsa ntchito, kuti muthe kupanga zinthu za SaaS, mapulogalamu a foni yam'manja, masewera ndi ntchito popanda ndalama zolipira kapena ndalama. Chotsani lisensi patsamba lililonse la model, chifukwa mamodeli ndi mawu ena sangakhale opanga malonda. Audio yomwe imapangidwa pa TTS.ai imatsatira ndondomeko yanu: kugwiritsa ntchito kwa munthu mmodzi pazowonjezera zaulere, kugwiritsa ntchito kwamalonda pa chilichonse cholipira.
5.0/5 (1)

Kodi tingachitire chiyani kuti tisinthe? Maganizo anu amatithandiza kuchotsa mavuto.

Pezani Open Source TTS Tsopano

20 + mapangidwe a otsegulira, onse ndi lisensi yawo yoperekedwa. Mutha kugwiritsa ntchito API yathu kapena kuyendera nokha, kusankha ndi kwanu.

Kulembetsa kwaulere Onani mtengo