Dictation software makes it easy to navigate your computer and communicate without typing a single phrase.
This flexibility is great if you simply need a break from your keyboard, but it’s especially important for people with language-processing disorders or physical disabilities. Firing off a quick text or typing a memo can be difficult—or even totally infeasible—if you have limited hand dexterity or chronic pain, but this kind of software can make such tasks a relative breeze.
But the technology behind dictation software (also called speech-to-text or voice-recognition software) has some faults. These apps have difficult learning curves, and the inherent bias that humans program into them means that their accuracy can vary, especially for people with various accents, sociolects and dialects like African American Vernacular English, or speech impediments. Still, for those able to work within the technology’s constraints, our picks are the best options available for many people who need assistance using a word-processing tool.
Apple’s Voice Control comes installed with macOS, iOS, and iPadOS, so it’s free to anyone who owns an Apple device. In our testing, it produced accurate transcriptions most of the time, especially for speakers with standard American accents. Competing tools from Google and Microsoft averaged 15 points lower than Apple’s software in our accuracy tests. Among our panel of testers, those with limited hand dexterity loved Voice Control’s assistive-technology features, which made it easy to navigate the OS and edit messages hands-free.
But while the experience that Voice Control provides was the best we found for Apple devices, it often misunderstood words or entire phrases spoken by testers with regional or other American accents or speech impediments such as stutters. Although such accuracy issues are expected for speech-recognition modeling that has historically relied on homogenous data sources, other tools (specifically, Nuance Dragon Professional v16, which is available only for Windows) performed slightly better in this regard. Apple’s tool may also lag slightly if you’re running multiple processor-intensive programs at once, which our panelists said slowed their productivity.
At $700, Nuance Dragon Professional v16 is the most expensive speech-recognition tool we’ve found, but it’s the best option for people who own Windows PCs. Professional v16 replaces our previous Windows PC pick, the now-discontinued Nuance Dragon Home 15. It offers added functionality for those working in finance, healthcare, and human services—and is probably overkill for most people. (If you need a free PC option, consider Windows Voice Recognition, but know it has significant flaws.)
Like its predecessor, Professional v16 involves a learning curve at first, but the Dragon tutorial does a great job of getting you started. Our panelist with language-processing disabilities said Dragon was one of the most accurate dictation options they tried, and the robust command features made it possible for them to quickly navigate their machine. Like our Apple pick, Dragon had trouble with various American dialects and international accents; it performed better for those testers with “neutral” American accents. It also struggled to eliminate all background noise, though you can mitigate such problems by using an external microphone or headset. Although Dragon produced the fastest transcriptions of any tool we tested, this wasn’t an unqualified positive: Half of our panelists said that they preferred slower real-time transcriptions to Dragon’s sentence-by-sentence transcription method because they found its longer pauses between sentences’ appearance on their screen to be distracting.