About a Year Since I Stopped Using a Keyboard | My Experience Trying Aqua Voice, Typeless, and Superwhisper

It has been about a year since I started doing almost all my typing—emails, first drafts of articles, Slack replies, and prompts for Claude Code—using voice input.

At first, I thought, "Voice input will just end up being a mess of spoken language that doesn't get cleaned up." But now, I feel like my daily text-based work time has decreased by 1 to 2 hours.

This article is a summary of my honest personal experience using these three tools, aimed at those who are wondering, "Which one should I actually buy?" Since there are plenty of spec comparison articles out there, I will focus here on how I chose my voice input AI.

Why I ended up trying all three

Honestly, I tried to narrow it down to one at first. Having two or three subscriptions is bad for my mental health.

However, after using them for a while, I realized that voice input AI has clear "suitable" and "unsuitable" use cases.

For example, when I want to fire off rapid prompts to Claude Code like "fix this here," if I choose the tool with the highest editing quality, I have to wait two seconds for the input, which breaks my train of thought. Conversely, when I want to write a long email by voice, if I choose the speed-priority tool, it outputs the text exactly as spoken, which takes time to revise.

The stress of continuing to use the "wrong tool" every day is overwhelmingly greater than the cost of running two subscriptions. That is my conclusion after three months of testing.

Aqua Voice | The star that creates an experience where "speed doesn't break your thinking"

The one I have been using the longest among the three is Aqua Voice. It is a cloud-based voice input tool created by a US startup, and its unique model, "Avalon," is trained not for "transcription" but for "editing." It doesn't just output what you said; it removes fillers (like "um" and "uh") and outputs the text with the intended meaning polished.

What makes it amazing

• Latency is about 450ms. The text appears the moment you stop speaking. The feeling of "waiting for it to finish typing" is almost zero.

• Deep Context is a screen recognition feature that reads open code, Slack threads, and quoted parts of emails. Misconversions of function names and proper nouns are dramatically reduced.

• In the Pro version, you can register up to 800 custom dictionary terms (as stated on the official plan page). It can learn company names, service names, and personal names all at once.

I will also write honestly about its weaknesses

• Not compatible with iOS / Android. On smartphones, you end up going back to the OS's native voice input.

• Requires cloud transmission. Depending on the industry, it may conflict with operational rules for handling confidential information.

• There is no one-time purchase plan, so the total cost adds up over the long term.

How I use it

I use it almost exclusively when I'm throwing prompts at Claude Code one after another, saying things like 'change this function in this file like this.' For someone who triggers voice input over 100 times a day, a 450ms response time makes a cumulative difference in fatigue that is on another level. The price is $8/month (annual contract). If you type text on your PC for more than 30 minutes every day, I feel this is an amount you can easily earn back.

Typeless | The 'Magic That Erases Retakes' Works for Long Drafts

Typeless is an AI voice input tool created by a team from Stanford University, and its concept is simply 'Speak, don't type.' It has an extreme focus on returning pre-formatted text.

What makes it different from the others?

• It doesn't just automatically remove fillers, it even handles self-correction recognition. If you say, 'Oh, wait, I want to write it like this,' only the correct version remains in the final output.

• Supports all platforms: Mac / Windows / iOS / Android / Web. Among the three tools, Typeless is the only one that allows voice input in the same environment on a smartphone.

• SOC 2 compliant and zero data retention, so it's easier to pass corporate adoption reviews.

• The free plan is generous at 8,000 words per week (it is clearly stated on the official pricing page that no credit card registration is required).

Honestly, these parts are a bit tricky

• It's $30 if paid monthly. It drops to the equivalent of $12/month if paid annually, but in a world where ChatGPT Pro and Claude Pro are $20, '$30 a month for voice input alone' is a high psychological hurdle.

• To use it on a smartphone, you need to grant 'full access permissions,' which may not be allowed on company-issued devices.

• It is slightly inferior to Aqua Voice in terms of speed.

How I use it

Long emails to clients, drafts of proposals, and first drafts of articles. For situations where I want to create text in one go that I can read through and send as is, I use Typeless for almost everything. Even if I just pour out spoken language, the output comes close to a polished business document. The 'erasing retakes' behavior really cuts down on editing time for writers.

Superwhisper | Achieving 'No Voice Data Sent to the Internet' Locally on Mac

Superwhisper is a Mac-centric tool based on OpenAI's 'Whisper' model, designed to make the most of the Apple Silicon Neural Engine (a Windows version is also in beta). Its biggest feature is that it can be run completely offline. Depending on your settings, voice data is not sent to the cloud at all, and processing is completed locally.

Why it's chosen

• Offline processing is only available in Superwhisper among the three tools. It can be used with peace of mind even for highly confidential work.

• You can choose from multiple models, including Whisper Large v3, and decide the balance between accuracy and speed yourself.

• The custom dictionary and replacement rules are powerful, allowing you to mechanically fix words that are misrecognized.

• And above all, there is a one-time lifetime plan for $249 (available on the official Superwhisper website). If you plan to use it for 3 to 4 years, it beats the subscription models.

• It comes with a 30-day money-back guarantee, so the exit cost is low if it doesn't work out for you.

Weaknesses

• Latency is 1–2.4 seconds, so it loses to Aqua Voice in terms of real-time performance.

• Because the UI is feature-rich, it is difficult to master everything on the first day.

• It requires macOS "Input Monitoring" permissions, which can make some users psychologically wary.

• The Windows version is in beta, and it is not compatible with Android. The ecosystem is Mac-centric.

Comparison Map by Perspective

When you line up the five axes to see "which one is actually better," a clear three-way standoff emerges.

• Speed (Latency): Aqua Voice ~450ms > Typeless ~1s > Superwhisper 1–2.4s

• Formatting Quality: Typeless > Aqua Voice > Superwhisper

• Total Cost over 5 Years: Superwhisper $249 < Aqua Voice $480 < Typeless $720

• Mobile Support: Typeless (all OS) > Superwhisper (iOS available) > Aqua Voice (PC only)

• Privacy: Superwhisper (offline capable) > Typeless (zero data retention/SOC 2) > Aqua Voice (cloud transmission involved)

The tool that fits you best depends on which axis is most important to you. Since there is no tool that is "#1 in everything," it is easier to choose if you first decide what is the biggest stress you want to eliminate in your work.

Which one to choose if you have to narrow it down to one

I will write down realistic recommendations for those who say, "I don't intend to subscribe to all three, I want to narrow it down to just one," categorized by use case.

• For those focused on speed, coding, and chat → Aqua Voice. The value of $8/month can definitely be recovered if you type text on your PC for more than 30 minutes every day.

• For those focused on long-form documents who also want to use it on their smartphone → Typeless. If you can accept $12/month assuming an annual contract, you will feel the difference in formatting quality within a month.

• For those handling confidential information or wanting a one-time purchase for long-term use → Superwhisper. The $249 lifetime license offers the best value if you plan to use it for over 3 years

• For those who just want to try out what voice input is all about → Typeless free plan (8,000 words per week). No credit card registration is required and you won't be charged automatically, making it the safest entry point

Summary

• Voice input AI is no longer a question of "whether to adopt it," but a must-have

• Aqua Voice, Typeless, and Superwhisper form a "rock-paper-scissors" dynamic where each has different strengths. You should choose based on your specific needs

• For speed, choose Aqua; for formatting, Typeless; for privacy, Superwhisper

• I recommend trying the free tier of Typeless for 1-2 weeks first to physically experience which aspect provides the most benefit for your work

• If you spend even one hour a day typing on a keyboard, any of these three tools will definitely provide benefits that outweigh the cost

Personally, I can no longer go back to the version of myself before using voice input AI. The subscription costs invested in voice input are negligible when considering the cost-effectiveness of the time saved.

I believe it is worth trying, especially for those who have never reconsidered keyboard input because it felt like such a given.