Dataset Sample API
Pull a sample of consented, PII-safe recordings from our contributor network — so your team can evaluate audio quality before we scope a full license.
Built for AI teams evaluating MyDeepTech datasets — audio is live today, with image and video sampling available on request. Each dataset has its own slug and its own dedicated API key, issued to your team on request.
GET /api/datasets/{dataset_slug}/samples
Host: {your_api_host}
X-Api-Key: {your_api_key}
→ audio_url, duration, language, country, prompt
From sample to full pipeline
This page documents our sample API — a capped, free way to evaluate data quality before you commit. The same evaluate-then-scope model applies across image, audio, and video.
Evaluate
Pull capped, PII-safe samples for the languages and formats you need — audio is live today, with image and video sampling available on request.
Scope
Tell us the volume, languages, and modality — image, audio, or video. We scope a dataset and a licence fee around your exact requirements.
Connect
Once scoped, we deliver the full dataset to your pipeline — or connect you directly to ours, depending on what fits your workflow.
One endpoint. One header.
GET a capped batch of samples for any dataset by passing that dataset's key in the request header.
GET https://{your_api_host}/api/datasets/{dataset_slug}/samples
X-Api-Key: {your_api_key}
Each call returns a capped batch of PII-safe samples. Audio datasets include an audio URL, duration, language, country, and a broad gender/age range; video datasets include a video URL, duration, resolution, language, and country. No contributor names, exact ages, or device fingerprints are ever included.
Try it with curl
curl "https://{your_api_host}/api/datasets/yoruba-speech-dataset/samples?limit=10" \
-H "X-Api-Key: {your_api_key}"
Replace {your_api_host} and {your_api_key} with the values we issue for your dataset, and {dataset_slug} with the one you were given. Hosts and keys are dataset-specific — please don't share them outside your evaluation team.
What's in each audio sample
Every record returned by the API is already PII-safe, ready to drop into an evaluation pipeline.
Audio URL
A direct link to the recording file.
Duration
Length of the recording, in seconds.
Language
The language spoken in the recording.
Country
The contributor's country.
Gender & age range
Broad ranges only — never an exact age or identity.
Prompt
The prompt text the contributor read aloud.
No contributor names, exact ages, or device fingerprints are ever included in API responses.
What's in each video sample
Video datasets return the same privacy-safe shape as audio, tailored to visual content — available on request as we onboard contributors per market.
Video URL
A direct link to the video file.
Duration
Length of the video, in seconds.
Resolution
Frame size of the video (e.g. 720p, 1080p).
Language
The language spoken in the video, if any.
Country
The contributor's country.
Gender & age range
Broad ranges only — never an exact age or identity.
No contributor names, exact ages, or device fingerprints are ever included in API responses.
Languages ready today
Each dataset has its own slug and its own dedicated API host and key, issued to your team on request. We source additional languages and datasets based on client demand.
Language
Locale
Dataset Slug
Sample Endpoint
Yoruba
yo-NG
yoruba-speech-dataset
/api/datasets/yoruba-speech-dataset/samples
Swahili
sw-KE
swahili-speech
/api/datasets/swahili-speech/samples
Hausa
ha-NG
hausa-speech
/api/datasets/hausa-speech/samples
Igbo
ig-NG
igbo-speech
/api/datasets/igbo-speech/samples
API hosts and keys are issued per dataset, per client — we don't publish them on this page. Looking for video? Video-sample datasets are available on request — tell us your target languages or markets and we'll scope coverage. Request access below and we'll send your key(s) directly, along with a short walkthrough of the response format if useful.
Ready to evaluate a dataset?
Tell us which languages you need. We'll send your dataset-specific API key and set up a walkthrough of the response format.
AI Data. Human Intelligence.