Collection guides
How do you collect Urdu speech data for ASR training?
Updated 2026-08-01 · 4 min read
Short answer
Collect Urdu speech data by fixing the corpus specification first — 250–1,000 hours of audio from 500–1,500 native speakers, split across Dakhini (Hyderabad), Lucknawi, Dehlvi dialects and balanced for gender, age and recording condition. Recruit in Uttar Pradesh, Telangana where the target varieties are actually spoken, record to one written protocol (16 kHz telephony or 48 kHz studio), transcribe in Perso-Arabic (Nastaliq) with a documented convention for spoken urdu and spoken hindi are largely mutually intelligible; the distinction is mainly lexical and orthographic. decide up front whether transcription is in nastaliq, devanagari, or both., and accept the corpus against measured WER and metadata completeness rather than studio hours consumed.
Key takeaways
- Urdu has roughly 51 million speakers across Uttar Pradesh, Telangana, Bihar; a corpus that samples only one state will not generalise.
- Budget 250–1,000 hours for a first production ASR corpus, and at least 500–1,500 distinct speakers to avoid speaker overfitting.
- The hardest part is not recording — it is recruitment, dialect quotas, consent and transcription consistency across cities.
Step 1 — Write the Urdu corpus specification
A specification is the contract everything else runs against. For Urdu it must state the hours or speaker count, the dialect split across Dakhini (Hyderabad), Lucknawi, Dehlvi, Bihari Urdu, the demographic quotas, the recording condition, the transcription convention and the acceptance criteria.
Where teams get this wrong is by specifying hours without specifying speakers. Two hundred hours from eighty speakers trains a model that recognises eighty voices. The speaker count is the variable that governs generalisation, and for Urdu we recommend 500–1,500 distinct participants.
Step 2 — Set quotas before recruiting
Quotas are set before the first session and tracked daily. Retrofitting a quota after 60% of collection is complete usually means discarding data, because the remaining pool cannot correct the imbalance.
| Quota dimension | Typical target | Reason |
|---|---|---|
| Speakers | 500–1,500 | Enough distinct voices that the model learns Urdu phonology rather than a handful of speakers |
| Gender | 50 / 50 | Pitch and formant range differ; unbalanced cohorts bias recognition |
| Age | 18–25: 30%, 26–40: 40%, 41–60: 30% | Older speakers keep conservative Urdu forms younger urban speakers have dropped |
| Region | Uttar Pradesh, Telangana, Bihar | Covers 5 recognised dialect varieties |
| Condition | Studio / quiet room / field / telephony | Match the acoustic profile of your deployment |
Step 3 — Design the Urdu prompt script
Scripted prompts must be phonetically balanced for Urdu, covering Shares most phonology with Hindi but adds Perso-Arabic phonemes (/q/, /x/, /ɣ/, /z/, /f/) that many speakers merge, Dakhini differs substantially from north Indian Urdu in lexicon, morphology and intonation, Register shifts between colloquial Hindustani and formal Urdu change the vocabulary distribution sharply in sufficient density. Generic translated English scripts produce corpora that miss exactly the contrasts an ASR model struggles with.
Alongside scripted material, collect spontaneous speech. Indian Urdu specifically, and Dakhini in particular, are absent from public data dominated by Pakistani Urdu broadcast speech. Spontaneous data is where the disfluencies, hesitations and natural prosody live, and models trained only on read speech degrade sharply on real users.
Step 4 — Recording protocol and capture chain
- 48 kHz / 24-bit studio capture where the deployment is app or device audio; 8 kHz narrowband captured over a real telephony path where the deployment is a contact centre
- Documented microphone and interface chain per studio so files from different cities are interchangeable
- Automated checks for clipping, DC offset, noise floor and silence ratio on ingest
- Speaker metadata recorded at session time — dialect, district, age band, gender, education, device
- Written consent in the speaker's own language, retained for audit and covering commercial model training
Step 5 — Transcription and annotation in Perso-Arabic (Nastaliq)
Transcription is where Urdu corpora most often fail acceptance. Right-to-left Nastaliq tooling errors and diacritic loss Merged phonemes transcribed by sound rather than by etymology, or vice versa, inconsistently Dakhini forms replaced with standard Urdu
Spoken Urdu and spoken Hindi are largely mutually intelligible; the distinction is mainly lexical and orthographic. Decide up front whether transcription is in Nastaliq, Devanagari, or both. Decide the convention in advance — native script throughout, Roman for embedded English, or a tagged hybrid — and publish it as a style guide with worked examples. Two-pass QA by a second native reviewer measures against that guide rather than against personal preference.
Step 6 — Acceptance and delivery
Acceptance should be measurable: transcript accuracy sampled per batch, metadata completeness at 100%, audio validation pass rate, and quota adherence within an agreed tolerance. Deliver in your ingest format — WAV plus JSON or TSV manifests, with speaker IDs preserved and a consent register attached.
Roll delivery in batches rather than one final handover. Batch delivery lets your team catch a format mismatch in week two instead of week ten, and lets training start before collection ends.
Recruitment reality in Urdu-speaking regions
Fix the script decision before fielding; retro-transcribing a Nastaliq dataset into Devanagari after delivery costs as much as the original transcription pass.
Our studio network covers Hyderabad, Lucknow, Delhi, which is what makes dialect quotas achievable without contracting a separate vendor per state.
Frequently asked questions
How many hours of Urdu speech data do I need for a usable ASR model?
250–1,000 hours is the usual first production corpus for Urdu, on top of any pretrained multilingual base. Fine-tuning an existing multilingual model can show measurable gains from 50–100 hours if the data matches your deployment acoustics.
How many speakers should a Urdu dataset have?
500–1,500 distinct native speakers. Speaker diversity matters more than raw hours once you are past the first hundred hours.
Which Urdu dialects should be covered?
At minimum Dakhini (Hyderabad), Lucknawi, Dehlvi. Which ones dominate your quota depends on where your users are, not on which dialect is considered standard.
How is code-mixing handled in Urdu transcripts?
Spoken Urdu and spoken Hindi are largely mutually intelligible; the distinction is mainly lexical and orthographic. Decide up front whether transcription is in Nastaliq, Devanagari, or both. We fix the convention in the style guide before collection and QA against it, because inconsistent code-mix handling is a common cause of silent WER inflation.
How long does a Urdu collection take?
A 100–300 hour Urdu programme typically runs 3–6 weeks from signed scope to final delivery, with rolling batches from week two.
Related reading
Turn this into a dataset specification
Tell us the languages, speaker count and minutes. You get a written scope, a protocol and a fixed price within one working day.