Nothing’s New Essential Voice AI Feature Has Changed How I Send Text Messages

Every smartphone brand in 2026 is going nuts over AI features, trying to find new ways to add them to their next product. Whether it is a tool that rewrites your texts in Shakespearean English or a magic eraser that removes your ex from every photo. All of them seem very pointless to me.

But one brand that is trying to do something practical with AI is Nothing. The company has just released its new AI-powered Essential Voice feature, which transcribes spoken words to text, and after testing it for the past couple of days, I think it has the potential to be a game-changer. 

What is Nothing Essential Voice? 

In case you don't know, Essential Voice is a new tool that allows users to input any text field with their voice. In simple terms, it is a speech-to-text feature that's built right into Nothing OS. You speak, and it converts your words into clean, ready-to-use text. But unlike regular tools, Essential Voice cleans up every stutter, including the 'umms', 'uhhs', and all the other half-finished words in between. 

The result is a polished version of what you intended to say, without even having to touch your keyboard. Accessing it is also easy, since you can bring it up from the keyboard, no matter which keyboard app you use, or long-press the Essential Key on the Phone (4a) Pro to trigger it from anywhere. That second option is the one I found myself using most. But let's take a look at what it can do. 

What Is It Like Using Nothing Essential Voice?

I typically struggle with speech-to-text tools since I have a thick Indian accent, so most of them don't work accurately. This is where the Essential Voice feature surprised me. It understood everything I spoke clearly, even when I knew I had mispronounced something, and corrected it in the final text. 

More impressively, it knows when I stammered or went back to correct myself mid-sentence and left out those moments from the transcription. This might seem small, but it makes a huge practical difference. I also used it when I was getting late for the office and had a bunch of Slack messages piling up.

Rather than typing out replies, I put my faith in Essential Voice to dictate my messages while grabbing my stuff, and every transcription was mostly accurate. Even with earbuds on, Nothing's Essential Voice was able to understand my muffled words accurately and got the message right. It does make mistakes occasionally, but in my testing, it would only slip up about once every ten tries. 

Essential Voice Maps Your Life and Languages

I also tested it with my colleagues who spoke in Bengali and Malayalam, and it held up well across both languages, converting their speech to accurate English text. Though you do need to mention at the beginning what language you need the speech to be translated into. It has support for over 100 languages, according to Nothing and can automatically detect which one is being spoken. 

Moreover, Nothing has also added something called Personal Mappings to Essential Voice. It automatically fills in contact details, provided you have mentioned them before in the Essential app. When I referred to "my email" during the testing phase, it automatically filled in the correct email address. 

Where Nothing Essential Voice Falls Short

I want to be fair about the gaps, because there are a couple.

The speech processing time is noticeable since it sends the recording to Nothing's servers and waits to get a response back. Depending on your internet speed and the length of your message, you are looking at 5–20 seconds of wait after you finish speaking before the text appears. 

I think on-device processing would fix this almost entirely, especially on high-end devices like the Nothing Phone (3), which has a powerful Snapdragon 8s Gen 4 processor. 

It also often picks up random words as Simplified Chinese, which feels kind of fishy. But Nothing clearly mentions that it is using Google's Gemini for processing. 

My last gripe is that it's not context-aware. You cannot open Gmail, speak a prompt, and have it draft a professional email for you. It will only transcribe what you actually say, word-for-word. Nothing has confirmed that it is working on bringing context awareness to Essential Voice. This will allow the feature to understand whether you are writing an email or a casual message and adjust its tone accordingly.

Nothing's Essential Voice Is Practical, Not Flashy

Two days is not a lot of time, and there are parts of this feature I could not test properly in that window. But even in that short period, I didn't brush it off like most AI features. I found myself reaching for it more often than I expected. It feels more natural to just say what you want to rather than typing it out, and this is why I feel that Essential Voice is a more practical AI tool than the rest. 

I like Nothing's approach towards such features, like the new breathing mode in Nothing OS 4.1 that I also tried out recently. It leans towards making things easier than generating random fluff. The Essential Voice will arrive on the Nothing Phone (3) first, followed by the Nothing Phone (4a) and the Phone (4a) sometime later.