Public documentation format allowlist; not a claim of native multimodal architecture.
Request format
POST JSON file_url with a direct public HTTP(S) audio link, plus optional language yo/ig/ha/pcm/en. Local paths and private-network addresses are rejected. Dashboard local upload is a separate workflow.Source for Request format (opens in a new tab)
Output formats
JSON transcript text with optional language/duration metadata. SDK unwraps result envelopes; an empty text string means no detected speech. This is a fixed task response, not arbitrary schema-constrained generation.Source for Output formats (opens in a new tab)
Idle engine may return 503 during startup. Prefer: respond-async-on-cold-start requests a 202 queued job_id instead; GET /v1/audio/transcriptions/jobs/{job_id} polls completion. Failed startup attempts are not billed.Source for Fallback behavior (opens in a new tab)
Task statuses
Pinned SDK handles queued, running, processing, completed and failed. A completed response must contain a response object; malformed/error responses are not valid transcripts.Source for Task statuses (opens in a new tab)
Result retrieval
SDK polls server retry_after, bounded to 1–30 seconds, with a 5-second fallback. Default client total wait is 40 minutes and allows up to five consecutive transient poll failures. These are SDK defaults, not audio limits, retention TTLs or latency guarantees.Source for Result retrieval (opens in a new tab)
Usage metering
0.5 credits or NGN 0.25 per second; 30 credits/NGN 15 per minute; 1,800 credits/NGN 900 per hour. Billed on uploaded audio duration, not file size.Source for Usage metering (opens in a new tab)
Published price
One-off credit bundles: NGN 5,000 buys 10,000 credits; NGN 10,000 buys 20,000; NGN 50,000 buys 110,000 including 10,000 bonus. Starter Gift grants 500 credits. Published NGN tariff equivalents use the standard schedule; bundle bonuses change effective credit acquisition cost. Do not assume a universal credit-to-currency conversion.Source for Published price (opens in a new tab)
Access & pricing
Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.
No model selector sent on this route in the reviewed contract.
Access
POST https://api.9jalingo.org/v1/audio/transcriptions; JSON file_url and optional language, X-API-Key with stt scope. Optional cold-start job polling uses /v1/audio/transcriptions/jobs/{job_id}.
Regions
Inference hosting region not established; company location does not prove processing residency.
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 16 Oct 2026
Category
Usage
Rate
Unit
Speech transcription
Uploaded audio duration1
0.5 credits
second of audio
Speech transcription
Uploaded audio duration1
NGN 0.25
second of audio
Rate conditions
Charged on actual uploaded audio duration. Rounding/minimum not specified. Provider equivalent rates are 30 credits/NGN 15 per minute and 1,800 credits/NGN 900 per hour. Failed engine startup attempts are not charged. Bundle bonus credits change effective acquisition cost.
Privacy permits processing submitted text/audio to provide services and using information to improve AI systems; it does not establish a blanket no-training guarantee. Retention is purpose-based without a fixed TTL, and users can request deletion. Hosting/payment/analytics providers may receive necessary data. The pricing FAQ says generated audio and voice profiles can be deleted, but its broad no-sharing language does not remove the privacy policy vendor disclosures. No exact inference residency, zero-retention contract or independent security certification 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.
9jaLingo API
Not verified
No native inference caching contract reviewed. Stored voice embeddings, response buffers, audio storage, engine warmup and job polling are not prompt-cache support.
How it works
Not collected
No native inference caching contract reviewed. Stored voice embeddings, response buffers, audio storage, engine warmup and job polling are not prompt-cache support.
Minimum reusable input
Not collected
No native inference caching contract reviewed. Stored voice embeddings, response buffers, audio storage, engine warmup and job polling are not prompt-cache support.
Lifetime & refresh
Not collected
No native inference caching contract reviewed. Stored voice embeddings, response buffers, audio storage, engine warmup and job polling are not prompt-cache support.
How to enable it
Not collected
No native inference caching contract reviewed. Stored voice embeddings, response buffers, audio storage, engine warmup and job polling are not prompt-cache support.
Usage fields
Not collected
No native inference caching contract reviewed. Stored voice embeddings, response buffers, audio storage, engine warmup and job polling are not prompt-cache support.
Billing rules
Not collected
No native inference caching contract reviewed. Stored voice embeddings, response buffers, audio storage, engine warmup and job polling are not prompt-cache support.
Limits & data boundaries
Not collected
No native inference caching contract reviewed. Stored voice embeddings, response buffers, audio storage, engine warmup and job polling are not prompt-cache support.
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 hosted speech service under proprietary platform terms. This is not a claim of private foundation weights: the provider also publishes Apache-2.0 Spark-TTS-derived artifacts, and exact hosted checkpoint equivalence is unverified.
