Indian and multilingual human data

Human data for production AI, sourced and verified against your requirement.

Tell us the language, use case, speaker profile, environment, rights and delivery requirements. NaraDatum checks suitable existing supply first, sources privately where needed, and scopes buyer-funded collection only when existing data cannot meet the brief.

Submit a requirement See how verification works

Find verified data where it exists. Build custom data when it does not.

Real users do not speak like scripts.

Production speech rarely behaves like a clean benchmark.

Users interrupt, self-correct, switch languages, use regional accents, give short replies and speak in imperfect environments. Data that excludes those behaviours can leave important deployment failure modes untested.

Sample transcript
A

"Bhai, delivery kab tak aayegi? Customer wait kar raha hai."

00:02Hinglish

B

"Panch minute mein pahunch jayega. Traffic mein thoda delay ho gaya tha."

Hindi00:06

Illustrative. NaraDatum-recorded specimen. Not licensable supply.

Speech structure matters

One-speaker and two-speaker audio are NaraDatum’s current speech focus. The required structure is matched to the model and deployment scenario rather than treated as interchangeable.

One speaker

Isolated speech and speaker-specific evaluation

Isolated prompts, intent phrases, read speech where the use case needs it, spontaneous monologue, and pronunciation or acoustic evaluation against a single voice.

Two speakers

Dialogue, voice agents and real interaction

Natural dialogue, customer and support interaction, interviewer and respondent flows, voice-agent evaluation, turn-taking, interruptions, corrections, code-switching, short answers and context changes.

Three or more

Assessed against the deployment need

For three-or-more-speaker requirements, NaraDatum can assess existing supply, sourcing or buyer-funded collection feasibility where the deployment need and demand justify the additional participant, diarisation and QA complexity.

Tell us the speaker structure you need →

Three routes, in this order

Every requirement is checked against them in sequence. Collection is the fallback, not the starting point.

  1. Route 1

    Reviewed existing supply

    Where suitable existing data clears review, NaraDatum can move towards licensing without commissioning unnecessary new collection. Licensing terms are agreed per dataset, and availability is confirmed against the requirement.

  2. Route 2

    Managed sourcing

    Where suitable data may exist outside anything NaraDatum has reviewed, we locate a supplier who holds it and review their identity, rights and provenance before anything is licensed.

  3. Route 3

    Buyer-funded custom collection

    When suitable existing data does not meet the requirement, NaraDatum scopes a buyer-funded custom collection. A collection may be NaraDatum-controlled or partner-managed. In a partner-managed collection, the approved partner operates the collection workflow to the agreed project requirements; NaraDatum reviews the evidence and delivery against the agreed brief. Partner controls are not presented as NaraDatum system controls.

How availability works

NaraDatum does not publish an open catalogue. Availability is confirmed against each requirement after rights, provenance, quality and commercial availability are reviewed. Where supply clears review, it can be shared privately against the brief, with evidence and controlled sample access. Every requirement is matched against existing supply and sourcing options before custom collection is considered.

See how it works Submit a requirement

For data suppliers

Hold human data you can lawfully license?

Suppliers sit in Routes 1 and 2. NaraDatum works with organisations that represent that they own relevant human datasets, or that they hold documented authority to license them, and that representation is reviewed before any dataset is offered.

Supplier-owned data remains supplier-owned unless a written agreement says otherwise. Supplier applications are not published as a public listing.

How supplying works →

Verification

What we check, and what stays declared.

Supplier approval is not dataset approval.

We review supplier identity and authority first. Each dataset is then assessed separately across provenance, rights, consent, metadata, quality control and intended-use fit.

Supplier approval does not make every dataset approved. Each dataset has to clear its own NaraDatum review before we treat it as buyer-ready.

Buyer-ready means the reviewed evidence supports the stated intended use for the dataset that was reviewed. It is not a warranty of fitness, not a guarantee of legal compliance, and not a statement that the dataset suits every model or every use.

We keep the underlying evidence states separate:

  • what the supplier declared
  • what documents NaraDatum reviewed
  • what NaraDatum independently tested
  • what still remains unverified

Verification splits those four states into six named classes.

If the evidence does not support the intended use, we don’t treat the dataset as buyer-ready.

Consent, where NaraDatum collects

For NaraDatum-controlled collection, consent records are linked to the relevant contributor and data, with permitted-use scope retained as part of the project record.

Consent record structureIllustrative structure, not a live record
{
  "consent_version": "per consent record",
  "granted_scopes": [
    "MODEL_TRAINING",
    "EVALUATION"
  ],
  "scope_retained_with_recording": true,
  "withdrawal": "tracked to the consent record"
}

How verification works →

Known limitations travel with every NaraDatum delivery.

The Dataset Passport records what was reviewed, what was established, what was tested and what remains unknown.

Sixteen limitations are published with the specimen, not disclosed on request. The specimen is an illustrative capability specimen and is not offered as licensable production supply.

See an illustrative Dataset Passport What a delivery contains
Dataset Passport extractIllustrative specimen, evidence classes stated separately
Audio integrity
Re-verified by NaraDatum on 9 August 2026, 6 of 6 checksums matched the manifest
Ownership
Internal record, not independently verified
Collection-time QA record
Not established

Indic and multilingual coverage, scoped to the requirement

NaraDatum can scope requirements across major Indic languages, Indian English, regional speech, multilingual combinations and other buyer-defined languages or variants where suitable sourcing or buyer-funded collection is feasible.

  • Hindi
  • Bengali
  • Marathi
  • Telugu
  • Tamil
  • Gujarati
  • Urdu
  • Kannada
  • Odia
  • Malayalam
  • Punjabi
  • Assamese
  • Indian English
  • Regional variants
  • Code-switched combinations

Examples include Hindi, Bengali, Marathi, Telugu, Tamil, Gujarati, Urdu, Kannada, Odia, Malayalam, Punjabi and Assamese, alongside Indian English, regional variants and code-switched combinations. This list is illustrative, not exhaustive. Other languages or combinations can be scoped against the buyer requirement. This is capability coverage, not confirmed inventory. Exact availability is established against the brief. Language capability does not mean licensable inventory is already on hand.

Language capability and availability →

Have a data requirement? Send us this.

If suitable data exists, we'll assess it first. If it needs sourcing, we'll tell you. If it needs to be built, we'll scope that separately.

Submit a requirement General enquiry
  1. 01Language / code-switching
  2. 02Model or use case
  3. 03Training / evaluation / benchmarking purpose
  4. 04Speaker structure
  5. 05Domain / scenario
  6. 06Environment / channel
  7. 07Approx volume
  8. 08Technical format
  9. 09Transcript / annotation needs
  10. 10Metadata needs
  11. 11Rights requirements
  12. 12Timeline