Kutumiza Zidziwitso za Zifukwa

TTS ya nthawi yoyenera

Kutumiza mawu-ku-kulankhula ndi sub-second-first-audio latency. Kupangidwa kwa ogwira ntchito za mawu ndi ntchito zogwira ntchito.

Tilibe mawu a TTS m'chilankhulo chanu. Tikuthandizeni kuwonjezera anu! Kugulitsa mawu anu

Malemba

Kutumiza
0/5,000 maonekedwe ~0.3s Audio yoyamba

Zosankha za mawu

Mamodeli otha kutsitsa kanemayo okha.

Kusintha kwa Live

Dinani Stream kuti muwerenge nthawi yoyamba yokhala ndi mavidiyo

Kutulutsa

Zina zomvetsera zidzasewera pano pamene zikulowa.

0:00
Chidutswa choyamba:
Kuphatikiza kwa ma chunks: 0
Nthawi yonse:

Momwe Streaming TTS Amagwira Ntchito

1. Kutumiza Text

POST malemba kuti / v1 / tts / stream / monga Server-Sent Events pemphero.

2. Model amapanga

Kokoro chunks malemba ndi kulenga audio sample-by-sample pa GPU.

3. Stream masamba

Base64-encoded WAV zidutswa kufika pa SSE ndi kuyamba kusewera mofulumira.

4. Listen Live

M'malo mwake, wogwiritsa ntchito amadziwa kuti mawu ayamba pansi pa masekondi angapo, ngakhale pa mawu oyamba.

Kugwiritsa ntchito Malamulo

Popeza sub-second latency imatsegula zosowa zatsopano.

Mabungwe a mawu

Bots zokambirana zomwe zimayankha mofulumira monga munthu angachite.

Kulemba mawu

Kutanthauzira ndi dub mtsinje pa nthawi yoyenera popanda buffering paunts.

Masewera

NPC uthenga kuti amatsutsana ndi osewera mfundo mofulumira, palibe pre-zooneka VO.

Kupezeka

Owerenga mazenera ndi zida zothandizira zomwe zimayamba kulankhula pamene wogwiritsa ntchito akudina.

Realtime TTS Maphunziro

Kuyambira kwaulere, kusinthidwa pamene mukufuna zambiri

Opanda pake
  • Kokoro streaming (mwaulere model)
  • 500 characters per generation
  • 10 ufulu ma streams / tsiku kwa wogwiritsa ntchito anayi
  • Sub-second first-audio latency
  • SSE kutsitsa pa HTTPS
Otchuka kwambiri
Kukhazikitsa Akaunti yaulere
  • 15,000 characters pa signup
  • 5,000 chars per stream
  • Chithunzi cha API cha kulowa kwa programmatic
  • Kubadwa kwa mbiri
  • Sichikhala ndi malire a tsiku lililonse
Kulembetsa kwaulere
Pro
  • MOSS-TTS-Realtime (kapena pamene akusangalala)
  • 100,000 chars pa stream
  • Kusintha kwa GPU
  • Kuphatikiza kwa Voice agent + Twilio
  • Kuwonjezera mtengo mipaka
Kusintha

Funso Lofunsidwa Kawirikawiri

Kusintha kwa mawu kukhala mawu m'nthawi yeniyeni kumasintha mawu kukhala mawu molingana ndi momwe amapangidwira, osati kuyembekezera kuti mawu onse akwaniritsidwe. Sankhani yoyamba ya mawu imafika m'masekondi angapo. Izi zimapangitsa kuti ikhale yoyenera kwa ogwira ntchito za mawu, kujambula, komanso ntchito zogwirizana zomwe zimakhudza kukana.

TTS yachikhalidwe imapanga fayilo ya audio yonse pambuyo potumiza chilichonse — mumawamirira, kenako mumamvetsera mawu onse pamodzi. TTS yanthawi yeniyeni imagwiritsa ntchito Server-Sent Events (SSE) kuti ipereke mawu ochepaudio monga momwe model imawapanga. Wogwiritsa ntchito amamvetsera kuyambira kwa mawu pafupifupi nthawi yomweyo, ngakhale pazinthu zowonjezera zopitilira.

Kokoro ndi backend yosasinthika — imapanga mawu pafupifupi 100x mofulumira kuposa nthawi yachidule pa GPU yamakono. Tikuphatikiza MOSS-TTS-Realtime ngati njira ina yopanda mtengo; ogwiritsa ntchito adzatha kusankha papemphero limodzi pamene lidzatumizidwa.

Kusintha kwa mafoni a Kokoro kumatha kukhala 300-800ms pa intaneti ya anthu. Kusintha kwa netiweki kumatha kukhala kovuta kwambiri. Patsambali, mudzawona nthawi yoyamba pa UI, kuti mudziwe nthawi yayitali bwanji yomwe ikufunika.

Oyang'anira mawu omwe amayankha molumikizana, kuyankha kwaulere kwa ma media oyenda, masewera olumikizana ndi NPCs, owerenga othandizira omwe amayamba kulankhula nthawi yomwe wogwiritsa ntchito aklikhera, ndi pulogalamu iliyonse yomwe ikufuna kuyembekezera maola awiri kapena atatu kuti awonetsetse kuti ma audio amawoneka opanda pake.

Yes. POST to https://api.tts.ai/v1/tts/stream/ with the same body as the regular /v1/tts/ endpoint. The response is an SSE stream of base64-encoded WAV chunks. The free tier supports 10 generations per day per anonymous user; authenticated users get the full per-account character allowance.

Kokoro imagwiritsa ntchito mawu oyamba kuphunzira ndipo siyimapanga ma clone. MOSS-TTS-Realtime (kapena pamenepo) imathandizira kujambula mawu osatha kuchokera pa 3-second reference. Kuti mupange mawu omaliza lero, kugwiritsa ntchito tsamba la /text-to-speech/ ndi Chatterbox kapena GPT-SoVITS — izi sizingatheke kujambula mawu koma zimapanga mawu osiyanasiyana.

Kulipira kwa mawu ndi kalembedwe kamodzi ngati TTS yomaliza yokhazikika. Kokoro ndi yopanda mtengo (1x mtengo). MOSS-TTS-Realtime idzagwira ntchito pa mtengo wokhazikika (2x mtengo) pamene ikugwirizana. Protokali yotumizira imawonjezera ndalama zowonjezera.

Ndikofunika - kugwirizana ndi kulumikizana kwa mafoni ndi webhook ya mawu a Twilio kuti mupereke mawu ochokera pafoni. Platform yathu ya mawu ya voice agent yakhala ikuchita izi kwa IVR ndi kulumikizana kwakunja. Kuyambira pamapeto mpaka pamapeto kwa latency pafoni ndi 1-2 masekondi kuphatikizapo STT ndi LLM yankho.

Ngati netiweki yanu imangotulutsa gawo limodzi panthawi yopita, wosewerayo adzapita patsogolo mofulumira kuposa kulephera. Kwa mapulogalamu omwe satha kumvetsera ma gaps, bwererani ku malo omaliza opanda ma stream, kapena gwiritsani ntchito 500ms ya ma audio panthawi yoyamba kusewera.
5.0/5 (1)

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

Stream Kulankhula mu Real Time

Free kwa 10 zaka zayamba tsiku. Sign up kuti unlock the malire mfundo zonse ndi API kulowa.