MiniMax: Speech 2.8 Turbo
Speech 2.8 Turbo — ovoz belgilari va ifodalilikni boshqarish bilan tabiiy nutqning tez sintezi.
Model haqida
Bu tavsif model imkoniyatlarini yoritadi. Ailoot da tanlangan usul uchun mavjud kirish, chiqish va parametrlar quyida keltirilgan.
Suhbat replikalarini tez sintez qilish
Speech 2.8 Turbo audioni tez shakllantirishni ovozning tabiiy o‘zgarishi bilan birlashtiradi. Ishlab chiqaruvchi Turbo’ni ravon suhbat jarangi uchun mo‘ljallaydi: qisqa replikalar, matnni ovozlashtirish va ovozli muloqot. Bu karta matnni qabul qiladi va nutq audiosini qaytaradi.
Kulgi, nafas va undovlar
Bu avlodning HD versiyasi kabi, Turbo Speech 2.8 ovoz belgilarini qo‘llab-quvvatlaydi. Matnga kulgi, yengil kulish, yo‘tal, tomoq qirish, nafas olish, nafas chiqarish, xo‘rsinish va boshqa qisqa ovozlar belgilarini kiritish mumkin. Bunday boshqariladigan notekislik replikani suhbatga o‘xshatadi: unda mazmunga bog‘liq nafas va pauzalar paydo bo‘ladi.
Ovoz belgisi matnning aniq joyiga tegishli. Buning uchun model ko‘zda tutgan qavsli belgilar, masalan, (laughs) yoki (breath) ishlatiladi. Effektning ixtiyoriy tavsifi qo‘llab-quvvatlanadigan belgining teng o‘rnini bosmaydi.
Ovoz va til
Speech 2.8 oilasi 40 tilni, jumladan, rus tilini va tayyor ovozlar bilan ishlashni qo‘llab-quvvatlaydi. Ishlab chiqaruvchi ovozning tanib bo‘ladigan fakturasi, nafas xususiyatlari va raqamli buzilishlarni kamaytirishni ta’kidlaydi. Klonlash imkoniyati keyinchalik sintezda ishlatilishi mumkin bo‘lgan ovozni alohida tayyorlashga tegishli.
MiniMax ovoz tanlash hamda sur’at, balandlik va ovoz pardasini boshqarishni ko‘zda tutadi. Tezlik 0,5–2 oralig‘ida, balandlik 0 dan yuqori va 10gacha, ovoz pardasi o‘zgarishi esa −12 dan 12 gacha belgilanadi. Hissiyotni matn mazmuniga ko‘ra model tanloviga qoldirish yoki qo‘lda ko‘rsatish mumkin: quvonch, qayg‘u, g‘azab, qo‘rquv, jirkanish, hayrat yoki xotirjamlik.
Talaffuzni almashtirish lug‘ati va fonetik yozuv bilan aniqlashtirish mumkin; unda xalqaro fonetik alifbo, xitoy pinyini va yaponcha o‘qish variantlari ko‘zda tutilgan. Pauzalar aytiladigan parchalar orasida 0,01–99,99 soniya oralig‘ida belgilanadi. Pauza davomiyligi verguldan keyin ikki xonagacha aniqlikda belgilanadi; markerlar aytiladigan parchalar orasiga qo‘yiladi va ketma-ket joylashtirilmaydi. Bunday sozlamalar ismlar, qisqartmalar va bir nechta talqinga ega so‘zlarning jarangini ovozlashtirish g‘oyasiga moslashtirishga imkon beradi.
Bosqichma-bosqich chiqish va audio formati
Sintez 10 000 belgidan kam matnni qabul qiladi. 3000 belgidan uzun matn uchun ishlab chiqaruvchi oqimli chiqishni tavsiya etadi; tayyor audioni to‘liq olish ham ko‘zda tutilgan. Umumiy sintez sozlamalari MP3, PCM, FLAC, WAV, Opus va G.711 μ-law variantlari, mono yoki stereo’ni o‘z ichiga oladi; aniq so‘rovning mavjud formatlar to‘plami Ailoot vositalari bilan belgilanadi.
Turbo tezligi tayyor replikaning qat’iy kechikishi va’dasi emas, model varianti xususiyatidir. Bu yerda ushbu so‘rovning sonli kechikishi belgilanmagan. Ishlab chiqaruvchining sanab o‘tilgan kengaytirilgan sozlamalari Ailoot orqali tanlangan murojaat usulida qo‘llab-quvvatlanishni talab qiladi.
Qo‘shimcha ovoz sozlamalari 8, 16, 22,05, 24, 32 va 44,1 kHz diskretlash chastotalarini o‘z ichiga oladi. MP3 uchun 32, 64, 128 yoki 256 kbit/s bitreyt ko‘zda tutilgan. Ovoz tusini o‘zgartirish va bitta effekt qo‘llash ham mumkin: keng makon aks-sadosi, zal aks-sadosi, telefon ovozi yoki robotlashtirilgan ovoz. Ovozni o‘zgartirish effektlari audioni to‘liq olishda MP3, WAV va FLAC uchun, oqimli chiqishda esa faqat MP3 uchun qo‘llab-quvvatlanadi. G.711 μ-law formatlari 8 kHz chastotadan foydalanadi. Bu imkoniyatlar MiniMax sintez sozlamalariga tegishli.
Ishlab chiqaruvchi nutqni matn bilan sinxronlashtirishni ko‘zda tutadi: gaplar, so‘zlar yoki bosqichma-bosqich chiqishda so‘zlar bo‘yicha vaqt belgilari. Bunday natija talaffuzni matn ko‘rsatilishi bilan moslashtirishga yordam beradi. Bu qo‘shimcha imkoniyat tegishli javob formatini qo‘llab-quvvatlashni talab qiladi.
Ailoot Ultra · Ovozlashtirish POST /v1/audio/speech minimax/speech-2.8-turbo
minimax/speech-2.8-turboBoshlash
Speech 2.8 Turbo — ovoz belgilari va ifodalilikni boshqarish bilan tabiiy nutqning tez sintezi.
- Qabul qilinadigan kirish ma’lumotlari
- Matn
- Natijalar
- Audio
Ailoot da mavjud · kirish ma’lumotlari va usullar
Ailoot Ultra · Ovozlashtirish POST /v1/audio/speech
- Kirish
text- Chiqish
audio- Chiqish formati
- mp3
| Maydon | Tur | Talab | Qiymatlar va cheklovlar | Standart |
|---|---|---|---|---|
Maydonmodel | Turstring | TalabMajburiy | Qiymatlar va cheklovlar minimax/speech-2.8-turbo | Standart— |
Maydoninput | Turstring | TalabMajburiy | Qiymatlar va cheklovlar Eng kam uzunlik: 1 Birlik: Unicode belgilari | Standart— |
Maydonvoice | Turstring | TalabMajburiy | Qiymatlar va cheklovlar English_expressive_narrator, English_radiant_girl, English_magnetic_voiced_man, English_compelling_lady1, English_Aussie_Bloke, English_captivating_female1, English_Upbeat_Woman, English_Trustworth_Man, English_CalmWoman, English_UpsetGirl, English_Gentle-voiced_man, English_Whispering_girl, English_Diligent_Man, English_Graceful_Lady, English_ReservedYoungMan, English_PlayfulGirl, English_ManWithDeepVoice, English_MaturePartner, English_FriendlyPerson, English_MatureBoss, English_Debator, English_LovelyGirl, English_Steadymentor, English_Deep-VoicedGentleman, English_Wiselady, English_CaptivatingStoryteller, English_DecentYoungMan, English_SentimentalLady, English_ImposingManner, English_SadTeen, English_PassionateWarrior, English_WiseScholar, English_Soft-spokenGirl, English_SereneWoman, English_ConfidentWoman, English_PatientMan, English_Comedian, English_BossyLeader, English_Strong-WilledBoy, English_StressedLady, English_AssertiveQueen, English_AnimeCharacter, English_Jovialman, English_WhimsicalGirl, English_Kind-heartedGirl | Standart— |
Maydonresponse_format | Turstring | TalabMajburiy | Qiymatlar va cheklovlar mp3 | Standart— |
Kirish materiallari birikmalari
text_to_speech · Majburiy: input
Usul parametrlari va hayot sikli
| Parametr | Parametr talablari |
|---|---|
Parametrcontent_type | Parametr talablariapplication/json |
Parametrrequired | Parametr talablarimodel, input, voice, response_format |
Parametrvoice_ids | Parametr talablariEnglish_expressive_narrator, English_radiant_girl, English_magnetic_voiced_man, English_compelling_lady1, English_Aussie_Bloke, English_captivating_female1, English_Upbeat_Woman, English_Trustworth_Man, English_CalmWoman, English_UpsetGirl, English_Gentle-voiced_man, English_Whispering_girl, English_Diligent_Man, English_Graceful_Lady, English_ReservedYoungMan, English_PlayfulGirl, English_ManWithDeepVoice, English_MaturePartner, English_FriendlyPerson, English_MatureBoss, English_Debator, English_LovelyGirl, English_Steadymentor, English_Deep-VoicedGentleman, English_Wiselady, English_CaptivatingStoryteller, English_DecentYoungMan, English_SentimentalLady, English_ImposingManner, English_SadTeen, English_PassionateWarrior, English_WiseScholar, English_Soft-spokenGirl, English_SereneWoman, English_ConfidentWoman, English_PatientMan, English_Comedian, English_BossyLeader, English_Strong-WilledBoy, English_StressedLady, English_AssertiveQueen, English_AnimeCharacter, English_Jovialman, English_WhimsicalGirl, English_Kind-heartedGirl |
Parametrresponse_format | Parametr talablarimp3 |
Parametrmax_input_characters | Parametr talablarinull |
Parametrother_options | Parametr talablarifalse |
Parametrinput | Parametr talablariaynan UTF-8 matni; normallashtirmasdan Unicode belgilarini sanash |
Parametrmaterial_types | Parametr talablaritext |
Hayot sikli: synchronous_result_then_owned_library
GET /v1/media/tasks/{task_id} · Amalda qaytarilgan task_id yoki media kutubxonasidagi yozuv
GET /v1/media/assets/{asset_id}/content
GET /v1/media/assets
GET /v1/media/assets/{asset_id}
Shu model profilidan aniq ovozni tanlang.
Kirish belgilari va audio soniyalari — turli hisoblash birliklari.
Tariflar
Tarif ma’lumotlari so‘rov imkoniyatlarini qo‘shmaydi. Qo‘llab-quvvatlanadigan parametrlar tanlangan usulda ko‘rsatilgan.
- Tezkor rejim
- Ko‘rsatilmagan Tezkor rejim tanlanganda
- Flex rejimi
- Ko‘rsatilmagan flex rejimi tanlanganda
- Maksimal fikrlash darajasi
- Ko‘rsatilmagan Maksimal fikrlash darajasi uchun standart (default) tuzatish qo‘llanganda
Ailoot tariflari
Sozlamalar bo‘yicha narxlar
Har bir to‘lovni ko‘rsatilgan shartlarga moslang. Har bir muqobil sozlamalar to‘plami uchun bitta tarifdan foydalaning.
| Sozlama | Tariflar jadvali | To‘lanadigan miqdor | Hisob birligi | To‘lov |
|---|---|---|---|---|
| SozlamaTokenlar uchun bazaviy tariflar | Tariflar jadvaliAsosiy tarif | To‘lanadigan miqdor— | Hisob birligi— | To‘lov Kirish · Kirish tokenlari — / 1K token Chiqish · Chiqish tokenlari — / 1K token Kirish · Keshdan o‘qish — / 1K token Kirish · Keshga yozish — / 1K token Kirish · Keshga yozish, 1 soat — / 1K token |
| SozlamaUltra | Tariflar jadvali— | To‘lanadigan miqdorHar bir Unicode kod nuqtasi, jumladan Han belgilari, bitta belgi hisoblanadi. | Hisob birligi1K kirish belgisi uchun | To‘lov Kirish · Kirish belgisi $0.12 1K kirish belgisi uchun |
Ma’lumot uchun tarif tafsilotlari
Modelning ma’lumot jadvallari usul tanlash, narx hisobi yoki to‘lovlarni o‘zgartirmaydi. Amaldagi to‘lovlar tanlangan Ailoot usuli tariflari bo‘yicha hisoblanadi.
MiniMax: Speech 2.8 Turbo
Sozlamalar bo‘yicha narxlar
| Sozlama | To‘lovni hisoblash shartlari | To‘lov |
|---|---|---|
| SozlamaMiniMax: Speech 2.8 Turbo · Standart sozlama | To‘lovni hisoblash shartlariStandart konfiguratsiya | To‘lov Kirish · Kirish belgilari 0,00012 US$ kirish belgisi uchun |
Ailoot so‘rovdan oldin taxminiy summani band qiladi. Natijadan keyin o‘lchangan sarf asosidagi yakuniy to‘lovni yechadi. Oqimli javob bo‘yicha yakuniy hisob-kitob yakuniy sarf yozuviga asoslanadi. Summani band qilish va to‘lovni hisoblash tartibi →
So‘rov namunalari
Ailoot Ultra · Ovozlashtirish
set -eu
: "${AILOOT_API_KEY:?Set your Ailoot API key}"
: "${AILOOT_VOICE_ID:?Choose an accepted voice ID from this exact model profile}"
export AILOOT_VOICE_ID
python3 - <<'PY_REQUEST'
import json,os
from pathlib import Path
p=json.loads('{"model": "minimax/speech-2.8-turbo", "input": "Good afternoon! This is a short voice take."}')
voice=os.environ["AILOOT_VOICE_ID"]
allowed=json.loads('["English_expressive_narrator", "English_radiant_girl", "English_magnetic_voiced_man", "English_compelling_lady1", "English_Aussie_Bloke", "English_captivating_female1", "English_Upbeat_Woman", "English_Trustworth_Man", "English_CalmWoman", "English_UpsetGirl", "English_Gentle-voiced_man", "English_Whispering_girl", "English_Diligent_Man", "English_Graceful_Lady", "English_ReservedYoungMan", "English_PlayfulGirl", "English_ManWithDeepVoice", "English_MaturePartner", "English_FriendlyPerson", "English_MatureBoss", "English_Debator", "English_LovelyGirl", "English_Steadymentor", "English_Deep-VoicedGentleman", "English_Wiselady", "English_CaptivatingStoryteller", "English_DecentYoungMan", "English_SentimentalLady", "English_ImposingManner", "English_SadTeen", "English_PassionateWarrior", "English_WiseScholar", "English_Soft-spokenGirl", "English_SereneWoman", "English_ConfidentWoman", "English_PatientMan", "English_Comedian", "English_BossyLeader", "English_Strong-WilledBoy", "English_StressedLady", "English_AssertiveQueen", "English_AnimeCharacter", "English_Jovialman", "English_WhimsicalGirl", "English_Kind-heartedGirl"]')
if voice not in allowed: raise SystemExit("Choose an accepted voice ID from this exact model profile")
p["voice"]=voice
p["response_format"]="mp3"
Path("request.json").write_text(json.dumps(p,ensure_ascii=False))
PY_REQUEST
curl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/audio/speech \
-H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \
--data-binary @request.json -D response-headers.txt -o speech.mp3
Natijalarni olish va saqlash
Ailoot Ultra · Ovozlashtirish
mp3
GET /v1/media/tasks/{task_id}GET /v1/media/assets/{asset_id}/contentGET /v1/media/assetsGET /v1/media/assets/{asset_id}
Boshqa kontentni o‘qish
set -eu
: "${AILOOT_API_KEY:?Set your Ailoot API key}"
curl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/media/assets \
-H "Authorization: Bearer $AILOOT_API_KEY" -o media-assets.json
set -eu
: "${AILOOT_API_KEY:?Set your Ailoot API key}"
: "${AILOOT_MEDIA_TASK_ID:?Set the actual ID returned by the response or media-library metadata}"
export AILOOT_MEDIA_TASK_ID
resource_id=$(python3 -c 'import os,urllib.parse;print(urllib.parse.quote(os.environ["AILOOT_MEDIA_TASK_ID"],safe=""))')
curl --fail-with-body --silent --show-error "https://api.ailoot.ai/v1/media/tasks/$resource_id" \
-H "Authorization: Bearer $AILOOT_API_KEY" -o media-task.json
python3 - <<'PY_RESULT'
import json
from pathlib import Path
print(json.dumps(json.loads(Path("media-task.json").read_text())))
PY_RESULT
set -eu
: "${AILOOT_API_KEY:?Set your Ailoot API key}"
: "${AILOOT_ASSET_ID:?Set the actual ID returned by the response or media-library metadata}"
export AILOOT_ASSET_ID
resource_id=$(python3 -c 'import os,urllib.parse;print(urllib.parse.quote(os.environ["AILOOT_ASSET_ID"],safe=""))')
curl --fail-with-body --silent --show-error "https://api.ailoot.ai/v1/media/assets/$resource_id" \
-H "Authorization: Bearer $AILOOT_API_KEY" -o media-asset.json
python3 - <<'PY_RESULT'
import json
from pathlib import Path
print(json.dumps(json.loads(Path("media-asset.json").read_text())))
PY_RESULT
Barcha tayyor natijalarni saqlash
Qaytarilgan har bir to‘liq javob, fayl va kontent qismini saqlang.
To‘liq ikkilik natija va sarlavhalarni saqlang. JSON yoki bir nechta natija qaytarilsa, har bir ma’lumot elementini saqlang va qaytarilgan har bir haqiqiy imzolangan resurs URL manzilidan faylni yuklab oling.
Asl javobni tekshirish kerak bo‘lsa
Ma’lumotlarni solishtirib tekshirish uchun dastlabki operation_id/X-Client-Request-ID ni saqlang va so‘rovni qayta yubormang; audio holatini davriy tekshirish uchun alohida yo‘nalish yo‘q.
Javob maydonlari
| Maydon | Tur |
|---|---|
Maydonbinary audio | Turbytes |
Maydonresponse headers | Turheaders |
Xatolar va keyingi qadamlar
| Javob | Shart | Keyingi qadam |
|---|---|---|
Javob400 · invalid_request_error | ShartMajburiy maydonlar, material ifodasi yoki parametr qiymati noto‘g‘ri | Keyingi qadamYangi yaratish so‘rovidan oldin ko‘rsatilgan maydonlarni tuzating. |
Javobserver_error | ShartQaytarilgan javob operatsiya yoki natijani tekshirish kerakligini bildiradi | Keyingi qadamDastlab qaytarilgan operatsiya/vazifa ID sini, so‘rov sarlavhalarini va javobni saqlang; aynan shu operatsiya uchun qo‘llab-quvvatlanadigan o‘qish so‘rovlarini davom ettiring. Yaratish so‘rovini qayta yubormang. |
Taxalluslar
Model taxalluslari
Ailoot Ultra minimax/speech-2.8-turbo
Usul yo‘llari taxalluslari
Yuqorida ko‘rsatilgan usul yo‘llaridan foydalaning.
Chaqirish uchun tegishli ID lar
FAQ · MiniMax: Speech 2.8 Turbo
- Qanday kirish ma’lumotlari va natijalar mavjud?
POST /v1/audio/speech·minimax/speech-2.8-turbo:text→audio- So‘rov qaysi maydonlardan foydalanadi?
POST /v1/audio/speech·minimax/speech-2.8-turbo:model,input,voice,response_format