{"id":60947,"date":"2026-10-15T11:13:41","date_gmt":"2026-10-15T02:13:41","guid":{"rendered":"https:\/\/greensun.com.vn\/?page_id=60947"},"modified":"2026-10-19T11:49:11","modified_gmt":"2026-10-19T04:49:11","slug":"ai-data-services","status":"publish","type":"page","link":"https:\/\/greensun.com.vn\/en\/services\/ai-data-services\/","title":{"rendered":"AI Data Services"},"content":{"rendered":"\r\n
\r\n
\"Two<\/figure>\r\n\r\n\r\n\r\n
\r\n

AI data services<\/p>\r\n\r\n\r\n\r\n

AI data collection and annotation in Southeast Asia<\/h1>\r\n\r\n\r\n\r\n

Speech, image, video and text data for AI, collected, labeled and quality-checked in Vietnam, Indonesia and Thailand, with project management in English or Japanese.<\/p>\r\n\r\n\r\n\r\n

\r\n
Request a quote<\/a><\/div>\r\n\r\n\r\n\r\n
All services<\/a><\/div>\r\n<\/div>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n
\r\n
\r\n
\r\n
\r\n
\r\n

Quality AI training data from Southeast Asia<\/h2>\r\n\r\n\r\n\r\n

Green Sun supports AI developers and data teams with the human side of AI data: participant recruitment, data collection, audio processing, transcription, annotation, validation and QA.<\/p>\r\n\r\n\r\n\r\n

Drawing on project experience in Vietnam, Indonesia and Thailand, we handle every step from recruiting local participants to processing, quality checks and delivery. Requirements, progress reports and issues are discussed in English or Japanese.<\/p>\r\n<\/div><\/div>\r\n<\/div>\r\n\r\n\r\n\r\n

\r\n
\r\n

At a glance<\/h3>\r\n\r\n\r\n\r\n
    \r\n
  • Track record:<\/strong> about 5,000 participants and about 1,000 hours of raw audio in one multi-country speech project<\/li>\r\n\r\n
  • Countries:<\/strong> Vietnam, Indonesia and Thailand; other Southeast Asian countries on request<\/li>\r\n\r\n
  • Data types:<\/strong> speech, text, image, video and human feedback<\/li>\r\n\r\n
  • Communication:<\/strong> English or Japanese<\/li>\r\n\r\n
  • Getting started:<\/strong> a pilot batch before full production<\/li>\r\n<\/ul>\r\n<\/div><\/div>\r\n<\/div>\r\n<\/div>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n
    \r\n
    \r\n
    \r\n
    \r\n

    Why it matters<\/p>\r\n\r\n\r\n\r\n

    A model is only as good as its data<\/h2>\r\n\r\n\r\n\r\n

    Why AI teams work with a language specialist for data.<\/p>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n

    \r\n
    \r\n

    01<\/p>\r\n\r\n\r\n\r\n

    Native speakers make data reliable<\/h3>\r\n\r\n\r\n\r\n

    Accents, dialects, honorifics and everyday phrasing are learned from native speakers. Data from fluent, well-briefed speakers reflects how people actually talk and write.<\/p>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n

    \r\n

    02<\/p>\r\n\r\n\r\n\r\n

    Less data exists for many Asian languages<\/h3>\r\n\r\n\r\n\r\n

    Vietnamese and Japanese have far less ready-made training data than English. Collecting and creating it yourself, with clear guidelines, closes that gap.<\/p>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n

    \r\n

    03<\/p>\r\n\r\n\r\n\r\n

    Evaluation needs human judgment<\/h3>\r\n\r\n\r\n\r\n

    Automatic scores miss tone, naturalness and cultural fit. Human reviewers compare answers, score them and flag invented facts before users see them.<\/p>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n

    \r\n
    \r\n
    \r\n
    \r\n

    What we do<\/p>\r\n\r\n\r\n\r\n

    Our AI data services<\/h2>\r\n\r\n\r\n\r\n

    Six services that cover the data pipeline, from collection to QA and human review.<\/p>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n

    \r\n
    \r\n
    <\/div><\/div>\r\n\r\n\r\n\r\n

    AI data collection<\/h3>\r\n\r\n\r\n\r\n

    Data for training and evaluating AI models, collected in Vietnam and across Southeast Asia, from participant recruitment to multi-country collection.<\/p>\r\n\r\n\r\n\r\n

      \r\n
    • Participant recruitment<\/li>\r\n\r\n
    • Text, image and video<\/li>\r\n\r\n
    • Surveys<\/li>\r\n<\/ul>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n
      \r\n
      <\/div><\/div>\r\n\r\n\r\n\r\n

      Speech data collection<\/h3>\r\n\r\n\r\n\r\n

      Speech data for speech recognition, speech synthesis and conversational AI. We recruit participants to your criteria, record, check the data and manage quality.<\/p>\r\n\r\n\r\n\r\n

        \r\n
      • Speech recording<\/li>\r\n\r\n
      • Speaker criteria<\/li>\r\n\r\n
      • Data checks<\/li>\r\n<\/ul>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n
        \r\n
        <\/div><\/div>\r\n\r\n\r\n\r\n

        Audio processing and transcription<\/h3>\r\n\r\n\r\n\r\n

        About 1,000 hours of raw audio processed for Vietnamese, Indonesian and Thai projects: audio checks, utterance segmentation, transcription and data cleanup.<\/p>\r\n\r\n\r\n\r\n

          \r\n
        • Segmentation<\/li>\r\n\r\n
        • Transcription<\/li>\r\n\r\n
        • Data cleanup<\/li>\r\n<\/ul>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n
          \r\n
          <\/div><\/div>\r\n\r\n\r\n\r\n

          Data annotation<\/h3>\r\n\r\n\r\n\r\n

          Annotation of AI training and evaluation data, run with your guidelines, your tools and your workflow.<\/p>\r\n\r\n\r\n\r\n

            \r\n
          • Speech and text<\/li>\r\n\r\n
          • Image and video<\/li>\r\n\r\n
          • Classification<\/li>\r\n<\/ul>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n
            \r\n
            <\/div><\/div>\r\n\r\n\r\n\r\n

            Data validation and QA<\/h3>\r\n\r\n\r\n\r\n

            Quality standards, guidelines and acceptance criteria are agreed before the project starts, then each batch is checked against them.<\/p>\r\n\r\n\r\n\r\n

              \r\n
            • Sampling QA<\/li>\r\n\r\n
            • Rework management<\/li>\r\n\r\n
            • Multi-stage checks<\/li>\r\n<\/ul>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n
              \r\n
              <\/div><\/div>\r\n\r\n\r\n\r\n

              Human-in-the-loop<\/h3>\r\n\r\n\r\n\r\n

              Human review and evaluation for tasks AI cannot judge on its own. We propose an approach after reviewing your specification.<\/p>\r\n\r\n\r\n\r\n

                \r\n
              • AI output review<\/li>\r\n\r\n
              • Human validation<\/li>\r\n\r\n
              • AI evaluation<\/li>\r\n<\/ul>\r\n<\/div><\/div>\r\n\r\n\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n
                \r\n
                \r\n
                \r\n
                \r\n

                Data types<\/p>\r\n\r\n\r\n\r\n

                Data we work with<\/h2>\r\n\r\n\r\n\r\n

                What we do for each kind of data, and what it is used for.<\/p>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n

                Data type<\/th>What we do<\/th>Typical uses<\/th><\/tr><\/thead>
                Audio \/ speech<\/td>Recruit speakers, record, segment and transcribe<\/td>Speech recognition, speech synthesis, conversational AI<\/td><\/tr>
                Text<\/td>Collect or create text, then annotate and validate<\/td>NLP, LLMs, generative AI, search AI<\/td><\/tr>
                Image<\/td>Collect images and annotate objects to your guidelines<\/td>Computer vision, image recognition<\/td><\/tr>
                Video<\/td>Collect video and annotate actions or objects<\/td>Action recognition, computer vision, robotics<\/td><\/tr>
                Human feedback<\/td>Review, validate and evaluate AI output<\/td>AI output review, validation, evaluation<\/td><\/tr><\/tbody><\/table><\/figure>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n
                \r\n
                \r\n
                \r\n
                \r\n

                Regions<\/p>\r\n\r\n\r\n\r\n

                Projects across Southeast Asia<\/h2>\r\n\r\n\r\n\r\n

                Green Sun has delivered AI data and corpus projects in Vietnam, Indonesia and Thailand.<\/p>\r\n<\/div><\/div>\r\n<\/div><\/div>\r\n\r\n\r\n\r\n

                \r\n
                \r\n

                Vietnam<\/h3>\r\n\r\n\r\n\r\n