Voice

Voice-over (text to speech)

Natural voice-over in 70+ languages including Hebrew, always on the newest model. Hebrew gets the voice director first (prepare, on by default): niqqud on every Hebrew word with English words left in Latin letters, an English delivery tag per sentence, a stability that fits — the words stay yours, the captions (words, text) stay plain. Then every Hebrew take is heard back word by word (the words at risk, every English word and number): a word that came out wrong is read again, you pay for the delivered take only, and verify in the answer says what was checked. Say the delivery in mood. The mp3 is hosted for you — use it in /v1/studio/* as voiceover.

Billing: Per character (Flash/Turbo half price) · Rate limit: 60/min

post/v1/ai/voice

Request

  • Base URL: https://apihub.co.il
  • URL: https://apihub.co.il/v1/ai/voice
  • Auth: HTTP bearer

Headers

Idempotency-Keystring

Replay-safe: the same execution is returned instead of charging twice

X-Tagstring

Group calls into one job (e.g. one video) — see GET /v1/usage?tag=…

Request body

moodstring

A word for the director on the delivery: "motivational, builds to a shout", "calm explainer", "funny, deadpan"

textstring required
model'auto' | 'eleven_multilingual_v2' | 'eleven_v3' | 'eleven_v4' | 'eleven_v4_turbo' | 'eleven_flash_v2_5' | 'eleven_turbo_v2_5'

"auto" (the default) is always the newest model, eleven_v4 — the most emotive, speaks Hebrew, follows audio tags like [excited]. Name an older id only for a call that depends on it; Hebrew text always runs on eleven_v4, whatever is named

speednumber
stylenumber
prepareboolean

Hebrew: the voice director reads the text first (on by default) — niqqud on every Hebrew word (English words, brands and acronyms stay in Latin letters), an English delivery tag per sentence ([softly, smiling], [firmly], [yelling furiously, out of breath]…), a stability that fits; the words never change. Text that already carries your own tags or niqqud is reviewed, not skipped: your tags stay where they are and the words get their niqqud (a word you vocalised keeps your points when they fit the sentence). false = read exactly what you send

quality'fast' | 'balanced' | 'best'

Kept for older calls and ignored: every voice-over runs on the newest model ("auto")

voice_idstring

GET /v1/ai/voices — a stock voice, or a voice you cloned (POST /v1/ai/voices/clone)

next_textstring

The text that comes right after — the voice ends this one leading into it

stabilitynumber
timestampsboolean

Also return every word with its start/end (seconds) — to cut the take at its pauses, place it exactly, and time captions without a transcript

language_codestring

Force a language, e.g. he

previous_textstring

The text read just before this one (the previous passage) — the voice continues its flow instead of starting fresh

similarity_boostnumber

Response

Completed

idstring

Execution id — stable across internal retries; use it with GET /v1/executions/{id}

status'pending' | 'processing' | 'completed' | 'partial' | 'failed' | 'cancelled' | 'timeout'
operationstring
tagstring nullable

Your X-Tag, if sent

created_atstring date-time
completed_atstring date-time nullable
outputobject

Operation result (see each operation). While a Studio render is processing: stage (queued · transcribing · downloading · preparing · cleaning_voice · mixing_audio · rendering · finishing), phase (the engine's: download · prepare · voice · audio · encode · qa · upload, or null), progress (a percent, 0–100) and progress_fraction (0–1), eta_seconds (seconds left, never below 0) and estimate_seconds

errorstring nullable

Changes

No recorded changes to this operation across all 2 revisions of this API.