Buy a Hindi telephony speech dataset

Speech captured over a real telephony path — a genuine call through the network, not studio audio downsampled afterwards to imitate one.

Scenario-driven calls between a caller and an agent or IVR flow, each leg recorded separately, across a deliberate spread of handsets and network conditions.

50–60% of recorded time is accepted. Dropped calls, network artefacts beyond tolerance and unusable legs are discarded. Real network conditions are the point of the style and also its main cost.

• 16-bit PCM WAV audio, captured at 48 kHz / 24-bit and delivered at your target rate
• Verbatim Devanagari transcripts with utterance-level timestamps
• Call metadata: duration, codec, handset class, carrier, packet-loss events
• Intent label per call and per turn
• DTMF and hold, transfer and barge-in events
• Agent and caller leg identifiers
• Per-speaker metadata: age band, gender, region, dialect, education band, device
• Consent records mapped to speaker IDs and retained for audit
• Speaker-disjoint train / dev / test splits

The failure modes this style is prone to: studio audio downsampled to 8 khz and passed off as telephony, which has none of the codec or packet-loss characteristics that matter

• Codec and sample rate verified per file, rejecting any studio audio that has been downsampled to fake a telephony path
• Echo and double-talk checked on both legs
• DTMF events verified against the call log
• Level normalisation applied per leg, since handset output levels vary far more than studio microphones do

Bands, not a price list. A signed specification converts them into one fixed figure, and staged delivery lets you start training before the full corpus lands.

Public Hindi corpora skew heavily towards read newspaper text from educated urban speakers in Delhi and NCR. Rural Bihar and eastern UP speech, elderly speakers, and low-literacy speakers reading prompts aloud are largely absent.

Urban Hindi speech is Hinglish in practice. Expect 15-40% English tokens in spontaneous speech: numbers, brands, technology terms, and whole clause switches. Any Hindi corpus that excludes English tokens will not match production traffic.

• Inconsistent Devanagari vs romanised spelling for the same English loan word
• Nukta characters (क़ ख़ ग़ ज़ फ़) applied inconsistently by transcribers
• Numerals: whether to write digits, Devanagari numerals, or spelled-out words must be fixed in the style guide up front
• Honorific verb forms create long agreement chains that annotators shorten unless the guide forbids it

• Perpetual, transferable licence covering model training, derivative models and commercial deployment
• Written informed consent from every speaker, retained and auditable
• Compliant with India's DPDP Act 2023 handling requirements and mapped to GDPR obligations for EU buyers
• No scraped, resold or re-licensed third-party audio in any delivery

• Send your specification or use the standard one above
• Receive a fixed scope, price and delivery schedule within one working day
• Optional paid pilot of 10-20 hours against the same spec
• Recruitment and recording begin on signature, with staged batches
• Acceptance testing per batch, then final delivery with the full manifest