Service
Speech data collection
Scripted, conversational and wake-word recordings from consented Khmer speakers across Cambodia, on the phones, cars and rooms your users actually use.
What's included
- Custom scripts, prompts and wake-word lists written by Khmer linguists
- Conversational and call-centre dialogues at 8 kHz or 16 kHz
- Speakers balanced by province, age band and gender
- Phone, mobile app, far-field and in-car recording set-ups
- Signed consent and full metadata for every speaker
Sample deliverable
Speaker and session metadata attached to every recording.
Illustrative example. Real samples come with your free pilot.
{
"session": "rec_2026_0412", "script": "banking_v2",
"speaker": { "id": "spk_0732", "gender": "F", "age_band": "35-44",
"province": "Siem Reap", "consent_id": "c_51e2…" },
"device": "Android · Galaxy A14", "environment": "market",
"snr_db": 18.4, "clips": 142, "duration_min": 31.6
}How it works
- 1
Design
We agree the script, speaker mix, devices and environments.
- 2
Recruit
Vetted speakers from our network sign a Khmer consent form.
- 3
Record
Guided sessions in our app or studio, checked for noise and clipping.
- 4
Deliver
Clean audio, metadata and consent records in your format.
Frequently asked questions
How many speakers can you recruit?
From a few dozen to several thousand, across Cambodia's provinces and the diaspora. Tell us the mix you need and we confirm it during the pilot.
Can you record in noisy, real-world places?
Yes. We record in markets, streets, cars and offices on purpose, and tag the noise level of every clip.
Do you transcribe the recordings too?
We can. Most clients combine collection with transcription so they receive model-ready data.
Related services
Start with a free pilot
We produce a small batch to your spec so you can check the quality before you commit.