Published task directory, not a guarantee that every plan enables every language or language pair.
Language coverage
39 published TRANSLATION language entries; plan entitlements vary. Provider labels Konkani with code kk and Maithili with mai despite the two-letter-code heading. Preserve provider codes and verify against authenticated lookup results.Source for Language coverage (opens in a new tab)
Request format
application/json fields: t_text, t_source, t_target, t_source_name, t_target_name. These are documented request fields; missing OpenAPI required arrays are not filled by inference.Source for Request format (opens in a new tab)
All five generation routes are asynchronous. HTTP 200 with statusmessage=success can mean queued (is_queued=1, process_id) or saved as a draft (is_draft=1, draft_id). Drafts do not run automatically and must be resubmitted after capacity clears. HTTP 200 can also carry statusmessage=failed. Inspect body state, not only HTTP status.Source for Task statuses (opens in a new tab)
Completion callbacks
Final results use POST /translate-text-result appended to the configured callback Base URL. JSON fields include process_id plus text, src_lang, target_lang, translated_text, total_chars, total_words. Retain the documented text_character_lenth spelling where used. Callback authentication, retry schedule and idempotency contract are not established.Source for Completion callbacks (opens in a new tab)
Result retrieval
The reviewed text-translation route documents final webhook delivery; no separate text-translation result polling path appears in this OpenAPI snapshot.Source for Result retrieval (opens in a new tab)
Developer API subscriptions: API Base $49/month, API Growth $249/month, API Business $1,299/month. These are whole-plan fees, not model-specific marginal prices. No per-token, character or audio overage tariff established; the page shows a dollar symbol without an explicit ISO currency code.Source for Published price (opens in a new tab)
API Base/Growth/Business fair-use allowances: 1.5M/5M/10M Translation/TTS tokens and 500/750/1,000 ASR minutes. Token accounting, sharing across services and exact API reset interval are not defined. These are plan allowances, not context windows or per-request output ceilings.Source for Billing scope (opens in a new tab)
Request quotas
Fair-use table lists 1,500+ concurrent translation tasks for API Base, Growth and Business. Base/Growth describe per-key throttling with HTTP 429; Business offers custom SLA/dedicated capacity. These are concurrency allowances, not requests per second.Source for Request quotas (opens in a new tab)
Documentation discrepancy
Pricing describes API Business as 33 languages while fair use says 35+ and task directories list 38–39. Web-plan language and custom-voice counts also differ. Web pricing mentions seven-day resets, but API billing is monthly; the API allowance reset interval is not established. Confirm entitlement for the chosen plan.Source for Documentation discrepancy (opens in a new tab)
Compare the pricing page and task-specific language directory; do not silently choose the larger entitlement.
Access & pricing
Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.
API task path and vr_id/agent_id_key are not native model checkpoint identifiers.
Access
POST https://api.neuraspeechai.com/v1/translate-text; accesstoken header, paid API entitlement and configured callback Base URL. Final text-translation results delivered by webhook.
Regions
Processing region not published; the operator’s corporate location does not establish inference residency.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
Terms reserve platform technology and AI models to Airo Global Software (P) Ltd. while users retain uploaded content. Voice recordings are described as processed solely for requested features, with settings-dependent temporary processing or project-history storage. Retention is purpose-based without a fixed TTL. Cloud, payment, analytics and support providers may receive data. No exact inference residency or independently verified zero-retention/compliance guarantee established.
Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.
01 / INITIAL INPUTProcess & retain
The system processes eligible input and retains reusable state.
02 / LATER REQUESTCheck available state
Reuse depends on matching input, session state and the cache policy.
03 / CACHE HITReuse & generate
Reusing state can reduce repeated work. Any charges follow the billing rules below.
NeuraSpeech API
Not verified
No model-specific inference cache contract. Drafts, job status, voice profiles and retained output files are different features.
How it works
Not collected
No model-specific inference cache contract. Drafts, job status, voice profiles and retained output files are different features.
Minimum reusable input
Not collected
No model-specific inference cache contract. Drafts, job status, voice profiles and retained output files are different features.
Lifetime & refresh
Not collected
No model-specific inference cache contract. Drafts, job status, voice profiles and retained output files are different features.
How to enable it
Not collected
No model-specific inference cache contract. Drafts, job status, voice profiles and retained output files are different features.
Usage fields
Not collected
No model-specific inference cache contract. Drafts, job status, voice profiles and retained output files are different features.
Billing rules
Not collected
No model-specific inference cache contract. Drafts, job status, voice profiles and retained output files are different features.
Limits & data boundaries
Not collected
No model-specific inference cache contract. Drafts, job status, voice profiles and retained output files are different features.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 16 Oct 2026
What the measurements say
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
Commercial managed inference workflow under proprietary service terms. Native checkpoint identities, architecture and weight availability are not established; this record does not claim an independently verified private foundation-model release.
