Japanese AI Pronunciation Evaluation Feature Released

DolphinAI, K.K. (Headquarters: Toshima-ku, Tokyo, CEO: Masahiro Asakura) is pleased to announce that, the launch of the Japan's first AI pronunciation evaluation feature for Japanese, following the existing English pronunciation evaluation API "DolphinSOE (Smart Oral Evaluation)". At the same time, we have also launched an Online Demo System that allows users to experience it without registration, fully supporting Japanese learners' pronunciation correction with the help of AI technology.

Background of Release

Due to the advancement of globalization, the number of Japanese language learners continues to increase worldwide. However, mastering the unique pronunciations of Japanese poses a significant challenge for foreign learners. In particular, understanding and practicing the accurate pronunciation of phonemes that do not exist in their native languages (e.g., "ら-line", "つ", "ん", etc.) is not easy.

DolphinAI, K.K. has developed an AI evaluation function to support the improvement of Japanese learners' pronunciation based on the speech AI technology cultivated through its English pronunciation evaluation API and its track record of a cumulative total of 5 billion calls globally and 3 million monthly active users. This will enable the provision of optimized feedback for each learner in Japanese education settings and e-learning services.

Three Features of Japanese AI Pronunciation Evaluation Function

1. Word Level Evaluation (Word Mode)

• Precise Evaluation at the Phoneme Level: Compares the pronunciation of each individual phoneme (e.g., "か", "き", "く", etc.) with the standard pronunciation and scores accuracy.

• Detection of Pronunciation Errors: Automatically detects omissions (dropouts), mispronunciations, and insertion of extra sounds, providing specific points for improvement.

• Amount of Supported Words: Covers thousands of frequently used words in Japanese learning; evaluation is possible by specifying words without prior learning.

DolphinSOE Japanese AI Pronunciation Evaluation Word Evaluation Image

2. Sentence Mode

• Fluency Assessment: Analyze the fluency of the entire sentence and the pronunciation of natural speech rate. Up to 60 seconds in duration
  
• Pause Determination: Evaluate the pauses in natural speech flow and between sentences
  
• Comprehensive Scoring: Multi-dimensional assessment from the perspectives of accuracy, fluency, and completeness

3. Chapter Mode

• Long Speech Support: Comprehensive evaluation of pronunciation across paragraphs and chapters composed of multiple sentences. Up to several minutes in duration
  
• Consistency Check: Detects stability in pronunciation within long speeches and identifies quality degradation due to fatigue
  
• Visualization of Learning Progress: Tracks improvement through continuous practice

DolphinSOE Japanese AI Pronunciation Evaluation Chapter Evaluation Image

Online Demo

• Instant Experience from Your Browser: Practice pronunciation on the spot using your microphone through a browser on your PC or smartphone.
  
• Real-Time Feedback: AI displays evaluation results within a few hundred milliseconds after speech.
  
• Diverse Practice Exercises: Offers sample words, sentences, and passages tailored to different levels from beginner to advanced.

Expected Use Cases

• Japanese Language Schools and Educational Institutions
  Utilized as a pronunciation check tool in classes, improving the efficiency of individual instruction and reducing the burden on teachers

• Online Japanese Learning Platforms
  API integration into e-learning services, pronunciation correction function in self-study apps

• Corporate Training for Foreign Employees
  Business Japanese pronunciation training, improving speaking quality in presentations and meetings

• J.TEST Practical Japanese Proficiency Test Preparation
  Practice assuming the speaking test, identifying and overcoming pronunciation weaknesses in advance

---

About the Author

About DolphinAI, K.K.

■ Company Overview
DolphinAI, K.K. is an AI company that focuses on the Japanese language as its core. It develops products and provides technical services in speech recognition, speech synthesis, and voice dialogue. Tailored for the Japanese market, the company has built a high-precision voice-interaction platform that serves sectors such as call centers, meetings, and education.

■ Service - DolphinVoice (Speech-Interactive SaaS Platform)
Official Website: https://dolphinvoice.ai/
Key Features: ASR (Japanese, Mandarin, English, Mandarin-English mixed, Japanese-English mixed), TTS (Japanese, Mandarin, English)
Usage: About 7,000 accumulated hours per day in call center and AI meeting minutes scenarios

■ Service - DolphinSOE (Pronunciation Evaluation SaaS Platform)
Official Website: https://soe.dolphin-ai.jp/
Key Features: AI pronunciation evaluation and correction API (for English, Japanese and Mandarin)
Usage: In the Japanese market, commercial use is expected to be around 2 million times per month by December 2025.

■ Security Framework

• Achieved information security (ISO/IEC 27001) certification and SOC 2 Type 1 report.
  
• Robust infrastructure hosted in Japan-based data centers (AWS, etc.).
  
• Compliant with Japan's Act on the Protection of Personal Information (APPI) and global standards such as SSL encryption.
  
• Details ️

■ Contact

️　(+81) 03-6161-7298

　voice.contact@dolphin-ai.jp

　https://dolphinvoice.ai/