ºÚÁÏÉçÇøÍø

Welcome to our June issue of the Accessibility Minute Newsletter! This newsletter is produced by the ºÚÁÏÉçÇøÍø Digital Accessibility Office (DAO) and covers one accessibility skill or topic per month. Please visit the DAO website to access past newsletters. As always, thank you for taking a minute (or two) to read.

Speech-to-Text

Have you ever used the dictation feature on your phone to send a text or relied on auto-captions during a Zoom meeting? Then you’ve already experienced the power of Speech-to-Text (STT)—also known as automatic speech recognition (ASR), which transforms spoken language into written text. STT increases access for individuals who are d/Deaf, hard of hearing, or have motor/dexterity disabilities. It’s also useful for multitaskers, note-takers, those who prefer speaking over typing, and anyone who wants to work more efficiently. Regardless of how it's used, Speech-to-Text (STT) makes it possible to interact with technology hands-free.

How Does Speech-to-Text Work?

STT systems rely on a few key components:

• Acoustic models interpret the sound waves of your voice to translate audio into meaningful text.
• Language models work alongside acoustic models to help predict and arrange words into logical sentences based on patterns in spoken language.
• Signal processing reduces background noise and improves audio clarity to extract meaningful content for higher accuracy transcripts.

Although STT systems are getting better at handling different accents, background noises, and real-life conversation patterns, like any digital tool, it has its challenges. STT isn’t perfect, especially in noisy environments or when someone speaks very quickly, quietly, mumbles, or uses lots of slang terms. Also, contextual understanding is limited within STT software, and unclear pronunciation or industry-specific terms (think medical or legal jargon) can all lead to errors and may require training within the system.

It’s important to remember that STT processes voice data, so be aware of STT privacy policies and how that information is stored or shared. Speech-to-Text can be transformative when used thoughtfully, but understanding its limitations and respecting privacy are essential to getting the most out of it.

For higher accuracy results when using STT, we recommend you:

• Use a decent microphone (built-in mics can be hit or miss).
• Speak clearly and at your natural pace.
• Reduce background noise.
• If possible, train the STT software or add specialized vocabulary.

Whether you use it occasionally or rely on it daily, Speech-to-Text is a powerful tool for enhancing digital access and enabling more inclusive communication, one word at a time.