Custom datasets

Build a Custom Speech Dataset

Need a dataset that doesn't already exist? Thailand Voices can design and produce custom speech datasets around your exact requirements.

  • Thailand-based
  • Professional Recording
  • Native Speakers
  • Custom Data Collection
  • Quality Controlled

How a custom dataset comes together

We start from your model's needs rather than a fixed product. If the specification is still open, we can help define it — speaker counts, hours, styles, environments, transcription depth and licensing scope all move the cost and the timeline.

  1. 1Requirements
  2. 2Speaker Recruitment
  3. 3Recording
  4. 4Transcription
  5. 5Annotation
  6. 6Quality Assurance
  7. 7Delivery

What you define

  • Language and variant
  • Number of speakers and demographic quotas
  • Speaking style and scenario design
  • Recording environment and devices
  • Audio format, sample rate, bit depth and channels
  • Transcription and annotation requirements
  • Metadata schema and file structure
  • QA thresholds and acceptance criteria
  • Licensing, territory and AI training rights

Pilot, scale, deliver

A pilot batch confirms conventions and quality expectations. Once signed off, collection scales with progress reporting, and the final dataset arrives packaged with manifests and documentation.

Tell us what data you need

Send your dataset requirements — language, speakers, hours, environment, transcription and licensing scope. If the specification is still open, our team can help define it.