The speech evaluation system provides multi-dimensional scoring for various Chinese and English question types, covering a range of scenarios from reading words and paragraphs to answering open-ended questions. The system accurately evaluates metrics such as pronunciation, fluency, and integrity, making it ideal for integration into education and language learning applications. All question types are supported by the Java, C++, Python, C#, Android, iOS, WeChat mini program, and Harmony SDKs.
Supported languages | Question type classification | Question type name | Features | Evaluation dimensions |
English | Adult read-aloud | English words | Scores English words or phrases with phoneme-level accuracy. | Feedback dimensions include word score, stress, and phoneme mastery. |
English sentences | Scores English sentences with word-level and phoneme-level accuracy. | Feedback dimensions include fluency, integrity, pronunciation score, omissions, repetitions, and sentence-final intonation. | ||
English paragraphs | Scores English paragraphs with sentence-level and word-level accuracy. | Feedback dimensions include fluency, integrity, pronunciation score, omissions, and repetitions. | ||
English phonemes | Scores phonemes and supports multiple International Phonetic Alphabet (IPA) standards, including British and American English. | Feedback dimensions include the phoneme score. | ||
Phonics | Connects English letters to their corresponding sounds. This feature simplifies complex pronunciations into regular patterns to help students improve their pronunciation accuracy. | Feedback dimensions include the total score and pronunciation score. | ||
English word correction | Corrects the pronunciation of English words. It provides feedback on omissions, insertions, and mispronunciations by comparing the user's pronunciation to the correct one. | Feedback dimensions include the overall word score and the phonetic transcription of the user's actual pronunciation. | ||
Child read-aloud | English words for children | Scores English words or phrases with phoneme-level accuracy and supports phonics scenarios. | This question type is designed for children under 10. The feedback dimensions are the same as those for the adult English words question type. | |
English sentences for children | Scores English sentences with word-level and phoneme-level accuracy. Feedback dimensions include more than a dozen metrics, such as fluency, integrity, pronunciation score, omissions, repetitions, and sentence-final intonation. | This question type is designed for children under 10. The feedback dimensions are the same as those for the adult English sentences question type. | ||
Semi-open | English multiple-choice | The user reads the text of an option. The system automatically determines whether the user read the correct option. Only one answer is correct. | Feedback dimensions include the overall score. | |
English extended-choice | The user reads the text of an option. The system automatically determines whether the user read the correct option. The user's answer can be slightly extended. | Feedback dimensions include the overall score. | ||
English sentence selection | You can set multiple sentences for this question type. The system evaluates and scores only the sentence that the user reads. | Feedback dimensions include the overall score, word scores, and the ordinal number. | ||
English Q&A | A scenario is described, followed by a related question. The user answers the question based on the scenario they heard. | Feedback dimensions include the overall score, fluency score, integrity score, and pronunciation score. | ||
Open | Spoken composition | The user tells a short story about an image or retells something they have heard, using their own words and creative interpretation. | Feedback dimensions include the overall score, fluency score, pronunciation score, and key points. | |
Chinese | Adult read-aloud | Chinese Pinyin | Scores Chinese Pinyin with accuracy at the initial and final level. | Feedback dimensions include the total score, fluency score, score for each Pinyin syllable, score for the initial and final of each syllable, and tone score. |
Chinese words and phrases | Scores Chinese words and phrases with accuracy at the initial and final level. | Feedback dimensions include the overall score, initial and final scores, tone detection, and tone score. | ||
Chinese sentences | Scores Chinese sentences with character-level and Pinyin-level accuracy. | Feedback dimensions include fluency, integrity, pronunciation score, omissions, and repetitions. | ||
Chinese paragraphs | Scores Chinese paragraphs with sentence-level and character-level accuracy. | Feedback dimensions include fluency, integrity, pronunciation score, omissions, and repetitions. | ||
Chinese poems | Scores poems with character-level and Pinyin-level accuracy. | Feedback dimensions include fluency, integrity, pronunciation score, omissions, and repetitions. | ||
Chinese recitation | Detects the audio stream and provides a real-time score for each character. For optimal real-time performance, use short texts, such as classical poems. You can set the scoring precision as needed. | Feedback dimensions include fluency score, pronunciation score, and integrity score. | ||
Semi-open | Chinese multiple-choice | The user reads the text of an option. The system automatically determines whether the user read the correct option. Only one answer is correct. | Feedback dimensions include the overall read-aloud score. | |
Chinese sentence selection | You can set multiple sentences for this question type. The system evaluates and scores only the sentence that the user reads. | Feedback dimensions include the overall score, character scores, and the ordinal number. |