in a recent In the white paper, Margery Mayer, former president of Scholastic Education, called 2022 the “Year of Speech Recognition” in education. She may be right: the surge in the adoption rate of educational technology developers in the first half of this year reflects the recognition that technology can not only create a more attractive learning experience for students, but can also completely change the practice of early literacy teaching.

In previous years, such a vision seemed far-fetched.But as EdSurge pointed out earlier, the science behind children’s speech recognition has begun to mature and can Educational application Aroused the interest of edtech Developer, Educators and Researchers Same.

Part of the reason for the increasing use of speech recognition in education is the availability of today’s technology specifically built to satisfy children’s voices and behaviors. Previous speech recognition systems modeled on adult voices and lacked the accuracy required for educational environments.Now the child-specific speech recognition that supports oral reading fluency tools is more accurate and effective, and it may provide me Already described Increase “teaching rewards” for children and their teachers.

These new speech learning tools may also solve fairness and prejudice issues. The speech recognition that supports them is constructed with diversity in mind, so all accents and dialects can be understood equally-thus democratizing access to educational resources and reducing the risk of implicit bias, for example, in observational assessments . However, perhaps most importantly, these solutions are “personalized and authentic” because they make use of the students’ most natural learning tool: their own voice.

Although 2022 may be the year of speech recognition in education, for most educators, families and students, the technology itself is relatively new, even if they have voice assistants or smart speakers in their homes.And, given the power of this technology, I look forward to more solutions, such as Amplify’s mClass Express It is important to enter the market so that educators and others understand how they work and how to best use them.

Recently, I collaborated with Amelia Kelly, the vice president of speech technology at SoapBox Lab, to create Glossary Help educators and educational technology developers to better familiarize themselves with speech recognition and make informed decisions about its use in educational environments. The following are some of the key terms that are particularly important and explain why these terms are important.

Artificial Intelligence (AI)

A system designed to perform tasks autonomously rather than specifically programmed by humans.

Why it matters: Artificial intelligence is increasingly used in educational products, and this trend will undoubtedly continue in the next few years.

Machine learning

A subset of AI used to train computers to process large amounts of data so that they can perform tasks automatically and on a large scale.

Why it matters: Machine learning algorithms will “learn” and “improve” with each experience, thereby improving the voice recognition function of educational tools that support voice.

Deep learning

A machine learning algorithm based on deep neural networks that requires a large amount of training data and has a multi-layer architecture that allows them to model complex behaviors such as human speech and language use.

Why it matters: Neural networks are widely used in speech recognition, image recognition and other pattern recognition problems, which are suitable for K-12 learning.

Speech technology

The general term for technologies that allow users to use their voices to interact with products, services, and platforms. The underlying technologies to achieve this goal are speech recognition (understanding human speech), speech synthesis (computer speaking aloud), natural language processing (reading and understanding human language) and machine translation (converting human speech from one language to another ).

Why it matters: In the K-12 edtech environment, speech technology—especially speech recognition—can power many use cases, enabling independent reading exercises, language learning, dyslexia screening, learning feedback, and summative and formative assessments.

Automatic speech recognition/speech recognition/speech to text

Allows digital devices to convert speech to text, making it easier for the device to understand the speaker’s intentions. Words or concepts in the text can trigger actions (for example, “turn off the lights”, “text my sister”).

Why it matters: Once the digital device has a child’s reading transcript, it can compare it with the scoring rubric to determine the fluency and comprehension of reading. It can also provide a time stamp for individual words, so that teachers can easily find a specific word or phrase read by the child, and then listen to it again. These systems can also return pronunciation “confidence” at the utterance, word, and even phoneme level.


An intentional process used to reduce or eliminate unexpected deviations in speech recognition. Artificial intelligence systems can reflect the prejudices of their creators, resulting in poor and often biased experiences for underrepresented users. Especially machine learning algorithms will make decisions based on trained data sets. If these data sets cannot represent different groups of people, deviations may occur.

Why it matters: A biased system can amplify and spread the deep-rooted biases held by the system designer, as well as the limitations of available data sets. The impact of this bias in practice, evaluation and screening platforms, and children’s learning tools can be catastrophic. For example, if a biased system cannot understand a child’s accent or dialect while reading, it may feed back to that child that they are a bad reader when in fact they read correctly. On the other hand, a fair system will provide fair and uncompromising feedback and data to promote education companies and platforms to support children’s learning journey.

Speech assessment

When children read aloud, they use speech recognition technology to listen, recognize, and evaluate the learning situation invisibly.

Why it matters: Speech assessment tools used in classrooms and remotely can provide data on pronunciation and fluency in oral reading. They can also be used to screen for learning challenges such as dyslexia. When used for power evaluation, the data provided by speech recognition technology can support and improve children’s educational outcomes and help determine the type and level of support provided by teachers.

Keyword detection

A function of the speech recognition engine that can recognize keywords and phrases in speech.

Why it matters: Keyword detection is particularly useful when analyzing children’s speech. It can identify search terms in audio files alone, in sentences, or through background noise. For example, a child may choose his or her favorite animal from the list. Keyword detection can score every possible response, triggering a response in a game or course.

Pronunciation assessment

Assess the pronunciation quality of a word or phrase.

Why it matters: Pronunciation assessment is a tool for teachers to save a lot of time, especially when supporting face-to-face observation assessment, because they provide teachers with scores, compare what the child actually said with the given target words, so that teachers can better understand Where students may be struggling and need more support or attention.

Fluency assessment

Assess the fluency of children’s oral reading.

Why it matters: Another tool for teachers to save time. When the child reads a passage, the speech recognition system will record and count the number of word substitutions, omissions, insertions, and correct words. In turn, this becomes a measure of fluency, sometimes expressed as “correct words per minute” or “WCPM”.

Speech Therapy Evaluation

A speech assessment that assesses speech patterns and sentence structure.

Why it matters: Speech recognition-driven screening and practice tools can identify speech patterns that may point to speech development disorders, enabling students to practice at home between speech therapy sessions, while also providing speech therapists with progress data.

Privacy design

A method of technology development, design, and process that can ensure that the privacy of individual users is protected from the earliest stage to the end user experience. Design privacy requires companies to maintain transparency when processing data. For example, they promise to only use the data they collect to improve services and not for any commercial purposes, such as resale, analysis, or advertising.

Why it matters: When it comes to children’s data rights, privacy cannot be considered after the fact or designed at a later stage. Privacy needs to be integrated into all levels of infrastructure, data, and processes, and has been part of the spirit and vision of solutions that support voice from the beginning.


Source link

Leave a Reply